1840 lines (1839 with data), 437.1 kB

{"timestamp": "2025-02-22T19:09:06.110227", "level": "INFO", "name": "__main__", "module": "data_preparation_v2", "function": "prepare_data", "line": 118, "message": "Loaded raw data from trials.csv with shape (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T19:09:06.284065", "level": "INFO", "name": "__main__", "module": "data_preparation_v2", "function": "prepare_data", "line": 131, "message": "Text Lengths -- Min: 9, Max: 5000, Median: 1406.0, Mean: 1795.21", "taskName": null}
{"timestamp": "2025-02-22T19:09:06.286548", "level": "INFO", "name": "__main__", "module": "data_preparation_v2", "function": "prepare_data", "line": 147, "message": "Data split: Train 1231 samples, Validation 264 samples, Test 264 samples.", "taskName": null}
{"timestamp": "2025-02-22T19:09:06.355051", "level": "INFO", "name": "__main__", "module": "data_preparation_v2", "function": "<module>", "line": 175, "message": "Prepared data saved to:\n - prepared_data/train_prepared.csv\n - prepared_data/val_prepared.csv\n - prepared_data/test_prepared.csv", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.174037", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.174516", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.174684", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.174824", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.174954", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.175069", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.176315", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.176497", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.176838", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.176876", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.176910", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.176942", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.176973", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177011", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177120", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177150", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177313", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177373", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177401", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177431", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177459", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177486", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177517", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177613", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177643", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177772", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177803", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177834", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177863", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177898", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.177929", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.178026", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.178058", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.208364", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.210797", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1407, Val: 176, Test: 176", "taskName": null}
{"timestamp": "2025-02-22T21:31:03.210925", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T21:31:06.862419", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:06.961510", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T21:31:10.861227", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T21:31:15.050973", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T21:31:15.510530", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:15.527236", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T21:31:16.015825", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T21:31:16.537548", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T21:31:16.961988", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T21:31:16.976533", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.430471", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.918584", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.929072", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.932589", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.933263", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.933326", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1407, 1017)", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.933357", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1407,)", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.933459", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.933491", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.933519", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.933604", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.933633", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-22T21:31:17.933661", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.828777", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.829191", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.829330", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.829454", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.829562", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.829672", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.830337", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.830463", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.830751", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.830783", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.830814", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.830845", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.830873", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.830905", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831004", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831033", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831187", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831247", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831275", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831305", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831332", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831359", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831394", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831485", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831513", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831636", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831665", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831698", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831725", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831754", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831782", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831869", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.831899", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.859875", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.862173", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 352, Test: 176", "taskName": null}
{"timestamp": "2025-02-22T21:47:42.862296", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T21:47:46.066429", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:46.153349", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T21:47:49.550073", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T21:47:53.211594", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T21:47:54.100947", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:54.128250", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T21:47:55.076818", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T21:47:56.101049", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T21:47:56.527720", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T21:47:56.542268", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T21:47:56.997033", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T21:47:57.485780", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T21:47:57.494632", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T21:47:57.498138", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T21:47:57.498708", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T21:47:57.498764", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T21:47:57.498795", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T21:47:57.498948", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T21:47:57.498982", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (352, 1017)", "taskName": null}
{"timestamp": "2025-02-22T21:47:57.499009", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (352,)", "taskName": null}
{"timestamp": "2025-02-22T21:47:57.499096", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T21:47:57.499135", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-22T21:47:57.499163", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.807491", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.807928", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.808076", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.808201", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.808313", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.808416", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.809226", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.809344", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.809623", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.809656", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.809689", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.809720", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.809752", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.809785", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.809887", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.809926", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810085", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810144", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810173", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810204", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810231", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810259", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810290", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810379", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810407", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810534", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810563", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810593", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810621", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810651", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810680", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810775", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.810805", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 225, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.839415", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.841762", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1495, Val: 176, Test: 88", "taskName": null}
{"timestamp": "2025-02-22T21:54:57.841904", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T21:55:01.694893", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T21:55:01.795040", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T21:55:05.877100", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T21:55:10.286973", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T21:55:10.716166", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T21:55:10.731472", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T21:55:11.189814", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T21:55:11.682328", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T21:55:11.923747", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T21:55:11.932526", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.190621", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.469158", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.473937", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.479583", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.480463", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.480523", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1495, 1017)", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.480553", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1495,)", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.480653", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.480685", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.480713", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.480773", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.480802", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (88, 1017)", "taskName": null}
{"timestamp": "2025-02-22T21:55:12.480832", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (88,)", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.494296", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.494829", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.494983", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.495108", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.495224", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.495540", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.496265", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.496401", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.496749", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.496797", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.496836", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.496872", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.496913", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.496950", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497060", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497120", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497300", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497371", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497404", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497445", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497481", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497513", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497544", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497638", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497677", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497808", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497841", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497873", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497915", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497945", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.497977", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.498070", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.498103", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.525922", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.528277", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1495, Val: 176, Test: 88", "taskName": null}
{"timestamp": "2025-02-22T22:07:48.528423", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:07:52.379910", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:07:52.480374", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:07:56.568767", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:08:00.980387", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:08:01.412023", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:01.427524", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:08:01.884487", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:08:02.377409", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:08:02.618496", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:02.627441", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:08:02.885070", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:08:03.162903", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:08:03.167348", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:08:03.171628", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:08:03.172339", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:08:03.172405", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1495, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:08:03.172437", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1495,)", "taskName": null}
{"timestamp": "2025-02-22T22:08:03.172539", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:08:03.172573", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:08:03.172604", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-22T22:08:03.172664", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:08:03.172692", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (88, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:08:03.172719", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (88,)", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.307398", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.307668", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.307800", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.307927", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.308066", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.308190", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.308790", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.308916", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309201", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309234", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309267", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309297", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309326", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309360", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309463", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309494", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309652", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309717", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309746", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309775", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309805", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309832", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309863", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309950", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.309978", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.310105", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.310137", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.310167", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.310196", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.310226", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.310255", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.310346", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.310377", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 193, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.339606", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.342758", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:08:21.342906", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:08:24.601597", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:24.686166", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:08:28.154383", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:08:31.885538", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:08:32.547673", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:32.569067", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:08:33.270376", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:08:34.029787", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:08:34.690660", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:08:34.711388", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:08:35.420855", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:08:36.178133", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:08:36.183502", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:08:36.186416", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:08:36.187082", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:08:36.187157", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:08:36.187189", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:08:36.187407", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:08:36.187441", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:08:36.187468", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:08:36.187578", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:08:36.187607", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:08:36.187634", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.897947", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.898578", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.898746", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.898876", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.898996", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.899111", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.899813", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.899942", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900242", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900279", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900319", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900350", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900385", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900419", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900525", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900554", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900719", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900788", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900818", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900850", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900878", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900906", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.900936", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.901035", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.901065", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.901209", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.901245", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.901277", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.901313", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.901344", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.901376", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.901474", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.901506", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.931080", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.933445", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:22:26.933560", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:22:30.470864", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:30.556102", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:22:34.284848", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:22:38.271809", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:22:38.979679", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:39.001390", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:22:39.752342", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:22:40.561207", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:22:41.268846", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:22:41.288964", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.050018", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.867061", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.871784", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.876135", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.877120", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.877208", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.877245", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.877418", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.877458", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.877491", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.877632", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.877665", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:22:42.877696", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.334843", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.335372", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.335524", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.335666", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.335794", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.335913", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.336634", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.336757", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337051", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337084", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337118", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337150", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337180", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337215", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337326", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337358", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337518", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337589", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337618", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337674", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337735", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337775", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337811", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337920", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.337958", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.338164", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.338215", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.338248", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.338279", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.338310", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.338341", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.338451", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.338482", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.368461", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.370806", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:25:04.370892", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:25:07.902451", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:07.985271", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:25:11.719225", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:25:11.827097", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:11.890320", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:11.957492", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:11.977264", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:12.003044", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:12.185677", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:12.241497", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:12.274142", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:12.315854", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:12.330694", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:12.486976", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:12.675569", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:12.834826", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:12.941757", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:12.953488", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.006009", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.060098", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.079616", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.198381", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.232005", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.268294", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.452086", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.539198", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.597834", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.625347", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.740224", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.808174", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.919442", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:13.931000", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.007290", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.058317", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.069657", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.157513", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.228711", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.321409", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.333499", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.361960", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.392643", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.443929", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.558139", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.603172", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.654825", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.713824", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.756425", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.794621", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.813696", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.834687", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.849796", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.863189", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.875554", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.889300", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.956829", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:14.985175", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:15.122856", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:15.172970", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:15.317874", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:15.555076", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:15.672898", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:15.710127", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:15.770518", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:15.851144", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:15.967200", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:15.987292", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:25:16.697537", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:16.717943", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:25:17.477119", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:25:17.531018", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:17.603306", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:17.620122", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:17.653059", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:17.708280", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:17.721235", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:17.783990", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:17.844899", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:18.093628", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:18.196854", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:18.214794", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:18.299484", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:18.321425", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:18.347587", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:25:19.058404", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:25:19.078448", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:25:19.831257", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:25:19.875762", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.016205", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.120636", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.225687", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.425110", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.508133", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.535018", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.552409", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.672731", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.677086", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.682550", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.683231", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.683288", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.683318", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.683450", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.683479", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.683507", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.683617", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.683643", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:25:20.683671", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.641449", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.641829", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.641973", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.642102", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.642802", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.642966", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.644071", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.644212", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.645157", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.645197", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.645298", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.645343", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.645590", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.645628", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.645756", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.645789", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.645981", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646055", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646088", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646123", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646153", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646184", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646213", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646310", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646340", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646491", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646522", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646553", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646583", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646613", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646643", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646737", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.646766", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.678824", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.681201", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:27:34.681507", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:27:38.234744", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:38.317376", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.093696", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.203962", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.266848", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.333694", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.361122", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.390683", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.574089", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.591620", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.637519", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.677190", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.729105", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.743244", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:42.902676", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:43.090069", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:43.256552", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:43.367393", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:43.381277", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:43.432932", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:43.488911", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:43.520025", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:43.641755", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:43.675841", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:43.714073", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:43.898267", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:43.990285", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.055947", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.085201", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.208500", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.281362", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.392117", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.407196", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.484776", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.535532", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.546872", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.635093", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.709283", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.806963", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.826104", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.861259", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.892415", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:44.945416", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.059048", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.105838", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.169448", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.229446", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.272234", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.317367", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.339392", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.364523", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.384301", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.399706", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.416976", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.437814", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.509242", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.537607", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.675861", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.726542", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.874293", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:45.933334", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:46.118909", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:46.237677", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:46.276104", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:46.337740", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:46.418569", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:46.543123", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:46.563084", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:27:47.271957", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:47.292530", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.044307", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.101812", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.185665", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.205516", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.239057", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.307335", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.320117", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.381617", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.444894", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.559641", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.701385", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.811643", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.835320", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.921846", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.943337", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:48.968262", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:27:49.677596", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:27:49.697347", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:27:50.449983", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:27:50.495150", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:50.634849", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:50.744350", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:50.848563", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.050835", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.133933", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.162176", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.179719", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.285928", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.301849", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.306190", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.309169", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.309856", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.309915", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.309945", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.310137", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.310168", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.310200", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.310313", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.310340", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:27:51.310368", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.089408", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.089961", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.090113", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.090239", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.090356", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.090466", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.091302", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.091436", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.091734", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.091772", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.091813", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.091846", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.091876", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.091908", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092013", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092171", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: lowercase", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092205", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092318", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092379", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092410", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092442", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092471", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092499", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092528", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092621", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092648", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: lowercase", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092679", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092804", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092832", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092861", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092888", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092917", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.092946", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.093031", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.093058", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: lowercase", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.093086", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.121249", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.123570", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:30:37.123692", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:30:40.637933", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:40.722452", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:30:44.430356", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:30:44.541271", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:44.603439", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:44.669374", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:44.696328", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:44.725851", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:44.907574", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:44.925215", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:44.971231", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:45.010328", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:45.061390", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:45.075319", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:45.233817", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:45.419036", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:45.583566", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:45.693877", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:45.707669", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:45.758967", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:45.814025", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:45.845105", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:45.966477", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.000501", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.038161", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.220917", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.312793", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.378154", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.407211", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.529623", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.601665", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.712149", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.727186", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.805000", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.860790", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.872161", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:46.959501", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.033496", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.131291", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.150428", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.185536", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.216415", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.269414", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.382595", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.429177", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.493222", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.552065", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.594320", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.639111", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.661006", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.686210", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.705881", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.721540", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.738438", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.758882", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.830180", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.858347", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:47.996148", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:48.045879", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:48.192127", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:48.250740", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:48.435279", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:48.553445", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:48.591040", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:48.652198", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:48.732721", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:48.856796", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:48.876341", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:30:49.587901", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:49.608756", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:30:50.359032", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:30:50.416372", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:50.498899", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:50.518691", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:50.551840", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:50.619155", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:50.631906", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:50.693016", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:50.755331", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:50.868400", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:51.008519", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:51.118009", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:51.142058", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:51.229123", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:51.250427", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:51.275326", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:30:51.982154", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:30:52.002317", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:30:52.750661", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:30:52.795878", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:52.935367", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.044470", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.147893", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.350187", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.433423", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.462027", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.479699", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.585752", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.601757", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.613455", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.617384", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.618064", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.618124", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.618156", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.618280", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.618310", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.618338", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.618450", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.618480", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:30:53.618507", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.079141", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.079753", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.079895", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.080014", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.080623", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.080795", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.081543", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.082292", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.082623", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.082661", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.082743", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.082777", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.082808", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083155", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083267", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083425", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: lowercase", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083457", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083569", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083635", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083664", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083692", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083722", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083751", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083781", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083885", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083912", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: lowercase", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.083943", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.084077", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.084105", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.084137", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.084164", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.084191", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.084219", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.084311", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.084341", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: lowercase", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.084369", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.115386", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.117802", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:33:56.117890", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:33:59.654874", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:33:59.738644", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:34:03.460913", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:34:07.920079", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:34:08.628845", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:34:08.649445", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:34:09.403248", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:34:10.318457", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:34:11.024662", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:34:11.045019", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:34:11.798941", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:34:12.647030", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:34:12.651617", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:34:12.657271", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:34:12.657991", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:34:12.658058", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:34:12.658092", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:34:12.658215", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:34:12.658246", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:34:12.658273", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:34:12.658381", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:34:12.658411", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:34:12.658440", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.160229", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.160818", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.160956", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.161070", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.161183", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.161290", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.162039", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.162173", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.162480", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.162514", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.162549", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.162581", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.162612", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.162646", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.162758", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.162789", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.162958", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163025", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163055", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163085", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163112", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163139", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163167", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163256", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163288", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163416", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163445", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163474", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163501", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163541", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163573", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163662", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.163691", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.194321", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.196924", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:36:43.197094", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:36:46.739851", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:46.823647", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:36:50.561227", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:36:55.009965", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:36:55.720245", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:55.741034", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:36:56.493374", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:36:57.410880", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:36:58.120413", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:36:58.140587", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:36:58.894499", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:36:59.744165", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:36:59.748388", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:36:59.753909", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:36:59.754577", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:36:59.754639", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:36:59.754671", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:36:59.754800", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:36:59.754832", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:36:59.754859", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:36:59.754970", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:36:59.755002", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:36:59.755030", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.611077", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.611661", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.611800", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.611922", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.612039", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.612153", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.612904", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.613098", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.613442", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.613482", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.613518", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.613553", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.613588", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.613708", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.613742", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.613927", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.613998", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614042", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614074", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614104", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614135", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614252", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614283", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614432", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614461", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614491", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614521", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614553", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614659", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.614688", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.644794", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.647258", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:39:20.647348", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:39:24.194695", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:24.278318", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:39:28.023435", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:39:32.316859", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:39:33.030631", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:33.051105", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:39:33.814607", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:39:34.714238", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:39:35.439019", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:39:35.459195", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:39:36.217332", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:39:37.032230", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:39:37.036716", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:39:37.041259", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:39:37.041974", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:39:37.042037", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:39:37.042068", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:39:37.042229", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:39:37.042263", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:39:37.042295", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:39:37.042408", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:39:37.042442", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:39:37.042472", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.114126", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.116570", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.116737", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.116871", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.117001", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.117123", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.118918", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.119068", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.119395", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.119433", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.119470", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.119506", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.119536", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.119657", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.119693", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.119869", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.119941", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.119971", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.120004", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.120035", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.120068", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.120172", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.120203", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.120348", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.120378", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.120411", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.120452", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.120481", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.120567", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.120601", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.150168", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.152744", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:41:55.152892", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:41:58.652395", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:41:58.736262", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:42:02.421589", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:42:06.245369", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:42:06.951865", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:42:06.972272", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:42:07.718267", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:42:08.510683", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:42:09.211426", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:42:09.231334", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:42:09.975698", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:42:10.704556", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:42:10.715126", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:42:10.717939", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:42:10.718455", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:42:10.718522", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:42:10.718566", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:42:10.718692", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:42:10.718727", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:42:10.718756", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:42:10.718885", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:42:10.718917", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:42:10.718948", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.379434", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.379913", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.380046", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.380164", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.380282", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.380391", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.381095", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.381222", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.381517", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.381560", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.381593", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.381623", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.381652", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.381683", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.381711", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.381869", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.381936", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.381967", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.382001", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.382034", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.382064", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.382095", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.382124", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.382254", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.382284", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.382313", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.382346", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.382375", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.382405", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.382435", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.411716", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.414502", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:44:17.414634", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:44:20.938095", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:21.022226", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:44:24.739804", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:44:25.772025", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:44:26.956087", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:44:27.949422", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:44:28.657137", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:28.677932", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:44:29.429390", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:44:30.092564", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:44:30.798470", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:44:30.818544", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:44:31.570912", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:44:32.181448", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:44:32.190922", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:44:32.194368", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:44:32.195059", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:44:32.195120", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:44:32.195152", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:44:32.195310", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:44:32.195348", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:44:32.195382", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:44:32.195507", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:44:32.195541", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:44:32.195574", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.012318", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.014275", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.014439", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.014639", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.014813", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.014958", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.016292", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.016448", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.016759", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.016794", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.016827", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.016858", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.016888", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.016921", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.016951", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017107", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017167", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017197", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017231", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017263", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017291", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017321", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017350", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017496", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017525", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017557", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017590", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017629", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017661", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.017691", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.046537", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.048830", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:53:43.048967", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:53:46.572237", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:46.656969", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:53:50.368582", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:53:51.401807", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:53:52.577617", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:53:53.575673", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:53:54.283139", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:54.303611", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:53:55.058630", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:53:55.726512", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:53:56.434239", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:53:56.454636", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.204904", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.819403", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.831588", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.834515", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.835076", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.835137", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.835175", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.835307", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.835343", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.835372", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.835483", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.835514", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:53:57.835543", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.257018", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.259636", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.260193", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.260340", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.260510", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.260685", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.262198", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.262380", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.262733", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.262773", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.262809", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.262843", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.262886", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.262932", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263109", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263175", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263206", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263242", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263273", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263304", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263336", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263496", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263528", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263560", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263591", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263620", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.263651", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.293505", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.296346", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:56:52.296465", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:56:55.849750", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:56:55.939047", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:56:59.706024", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:57:00.570743", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:57:01.535232", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:57:02.348965", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:57:03.061133", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:57:03.081841", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:57:03.833639", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:57:04.380956", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:57:05.102928", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:57:05.123175", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:57:05.888080", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:57:06.401296", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:57:06.411184", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:57:06.414371", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:57:06.415052", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:57:06.415115", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:57:06.415151", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:57:06.415284", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:57:06.415319", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:57:06.415350", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:57:06.415468", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:57:06.415503", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:57:06.415536", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.563076", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.563747", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.563884", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.564009", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.564127", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.564232", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.564989", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.565181", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.565560", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.565601", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.565643", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.565681", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.565714", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.565840", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.565948", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.565982", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.566014", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.566052", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.566085", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.566231", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.566259", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.566288", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.566320", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.566350", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.596836", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.599396", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T22:58:54.599524", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:58:58.146863", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:58:58.229948", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:59:01.972507", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:59:02.781364", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:59:03.681191", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-22T22:59:04.441167", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:59:05.154904", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:59:05.175959", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:59:05.932605", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:59:06.447806", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T22:59:07.158725", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T22:59:07.178791", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T22:59:07.934820", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T22:59:08.404031", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T22:59:08.411740", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:59:08.414798", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T22:59:08.415575", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:59:08.415641", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:59:08.415675", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T22:59:08.415847", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:59:08.415878", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:59:08.415907", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T22:59:08.416018", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T22:59:08.416047", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T22:59:08.416076", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.585092", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.585827", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.586003", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.586143", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.586268", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.586390", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.587142", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.587294", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.587601", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.587638", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.587679", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.587711", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.587742", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.587774", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.587803", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.587959", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588028", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588059", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588093", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588123", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588152", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588183", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588212", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588347", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588374", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588402", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588432", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588462", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588490", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.588518", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.618474", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.620892", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T23:01:46.620984", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:01:50.146284", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:50.232277", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:01:53.952985", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:01:56.176975", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 17000 characters", "taskName": null}
{"timestamp": "2025-02-22T23:01:57.173787", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:01:57.882879", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:01:57.903529", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:01:58.654578", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:01:59.324064", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:02:00.033062", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:02:00.052751", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:02:00.813094", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:02:01.430075", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T23:02:01.434778", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:02:01.440544", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:02:01.441242", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:02:01.441306", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:02:01.441339", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T23:02:01.441460", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:02:01.441492", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:02:01.441523", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:02:01.441636", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:02:01.441666", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:02:01.441694", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.980910", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.981643", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.981816", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.981942", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.982087", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.982216", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.983020", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.983148", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.983584", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.983671", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.983713", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.983748", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.983780", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.983817", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.983848", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984116", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984204", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984234", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984266", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984297", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984329", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984362", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984390", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984541", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984572", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984600", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984631", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984662", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984692", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:53.984724", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:54.014601", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T23:03:54.017052", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T23:03:54.017145", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:03:57.603078", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:03:57.687170", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:04:01.434497", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:04:03.666914", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 17000 characters", "taskName": null}
{"timestamp": "2025-02-22T23:04:04.667555", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:04:05.380219", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:04:05.400892", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:04:06.163218", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:04:06.832794", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:04:07.548985", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:04:07.569773", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.342061", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.977284", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.981844", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.985967", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.986532", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.986613", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.986643", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.986783", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.986814", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.986845", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.986976", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.987008", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:04:08.987037", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.052035", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.053600", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.053781", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.053925", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.054060", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.054196", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.054979", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055116", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055434", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055468", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055503", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055533", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055562", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055593", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055625", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055776", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055839", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055874", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055905", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055933", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055960", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.055990", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.056020", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.056154", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.056184", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.056215", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.056247", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.056280", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.056628", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.056686", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.087163", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.089526", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1407, Val: 176, Test: 176", "taskName": null}
{"timestamp": "2025-02-22T23:05:57.089614", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:06:01.097248", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:06:01.190051", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:06:05.427306", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:06:08.190893", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 17000 characters", "taskName": null}
{"timestamp": "2025-02-22T23:06:09.077092", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:06:09.574453", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:06:09.589993", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:06:10.118642", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:06:10.580951", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.047583", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.062061", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.557848", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.959034", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.964823", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.970390", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.971120", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.971191", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1407, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.971223", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1407,)", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.971357", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.971389", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.971425", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.971519", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.971548", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:06:11.971576", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.852923", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.854016", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.854152", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.854289", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.854404", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.854523", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.855300", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.855444", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.855746", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.855785", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.855819", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.855852", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.855891", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.855925", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.855955", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856122", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856183", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856214", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856244", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856273", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856300", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856328", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856355", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856484", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856512", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856542", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856571", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856600", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856632", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.856662", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.887360", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.889728", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1407, Val: 176, Test: 176", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.889856", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:26:16.890933", "level": "ERROR", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 267, "message": "Error in preprocessing pipeline: bad escape (end of pattern) at position 1", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.619345", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.620991", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.621163", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.621303", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.621431", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.621549", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.622144", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.622287", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.622610", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.622644", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.622676", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.622723", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.622752", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.622784", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.622812", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.622983", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623051", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623080", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623114", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623142", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623171", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623200", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623230", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623360", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623391", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623427", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623456", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623484", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623514", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.623545", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.653319", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.655691", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1407, Val: 176, Test: 176", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.655831", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:29:02.656745", "level": "ERROR", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 267, "message": "Error in preprocessing pipeline: bad escape (end of pattern) at position 1", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.674231", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.674775", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.674901", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.675016", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.675133", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.675250", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.675808", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.675921", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676203", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676243", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676279", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676311", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676340", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676376", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676408", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676559", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676621", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676650", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676681", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676710", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676737", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676765", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676792", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676914", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676941", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.676970", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.677005", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.677033", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.677061", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.677089", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.704790", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.707292", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1407, Val: 176, Test: 176", "taskName": null}
{"timestamp": "2025-02-22T23:30:35.707380", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:30:39.748300", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:39.848664", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:30:44.177221", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:30:47.007537", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 17000 characters", "taskName": null}
{"timestamp": "2025-02-22T23:30:47.916005", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:30:48.419091", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:48.435529", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:30:48.965617", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:30:49.451964", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:30:49.923723", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:30:49.938519", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.443719", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.855432", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.862714", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.867563", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.868526", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.868581", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1407, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.868613", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1407,)", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.868720", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.868751", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.868785", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.868902", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.868930", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:30:50.868958", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.253754", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.254554", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.254711", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.254849", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.254969", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.255088", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.255837", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.255971", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256270", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256304", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256338", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256370", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256400", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256433", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256537", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256568", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256728", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256792", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256822", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256854", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256883", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256912", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.256942", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.257034", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.257065", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.257204", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.257233", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.257262", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.257289", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.257317", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.257345", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.257432", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.257460", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.288083", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.290519", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T23:34:46.290611", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:34:49.867039", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:49.957036", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:34:53.732593", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:34:58.238237", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:34:58.964963", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:34:58.987889", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:34:59.753012", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:35:00.687579", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:35:01.410285", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:35:01.431320", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:35:02.198550", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:35:03.062143", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T23:35:03.067426", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:35:03.071922", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:35:03.072703", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:35:03.072768", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:35:03.072802", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T23:35:03.072944", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:35:03.072976", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:35:03.073004", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:35:03.073114", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:35:03.073144", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:35:03.073172", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.299021", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.299861", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.300016", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.300279", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.300470", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.300612", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.301379", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.301619", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302066", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302110", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302146", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302178", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302209", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302244", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302363", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302398", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302584", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302648", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302677", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302706", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302734", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302761", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302791", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302888", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.302920", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.303074", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.303104", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.303133", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.303161", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.303198", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.303229", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.303321", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.303355", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.333065", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.335822", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T23:36:36.335935", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:36:39.952058", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:40.042442", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:36:43.851835", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:36:48.400474", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:36:49.130035", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:49.153485", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:36:49.925161", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:36:50.864740", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:36:51.592652", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:36:51.614060", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:36:52.384767", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:36:53.252745", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T23:36:53.257492", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:36:53.263057", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:36:53.263877", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:36:53.263946", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:36:53.263978", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T23:36:53.264103", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:36:53.264137", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:36:53.264165", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:36:53.264276", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:36:53.264305", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:36:53.264334", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.435103", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.436124", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.436284", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.436417", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.436542", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.436656", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.437410", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.437572", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.437881", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.437916", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.437950", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.437983", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438014", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438047", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438170", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438203", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438395", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438466", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438497", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438529", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438558", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438587", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438617", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438725", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438754", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438900", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438931", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438962", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.438991", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.439021", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.439051", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.439150", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.439218", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.469523", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.471877", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T23:51:11.471964", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:51:15.037865", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:15.128120", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:51:18.869607", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:51:23.336955", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:51:24.052999", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:24.074149", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:51:24.828537", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:51:25.754574", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:51:26.464661", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:51:26.485363", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:51:27.243607", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:51:28.099774", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T23:51:28.104473", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:51:28.110032", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:51:28.110767", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:51:28.110825", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:51:28.110854", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T23:51:28.111025", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:51:28.111060", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:51:28.111088", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:51:28.111199", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:51:28.111229", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:51:28.111261", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.525266", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.526088", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.526243", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.526392", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.526531", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.526649", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.527966", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528137", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528459", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528498", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528544", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528578", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528609", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528644", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528675", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528838", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528909", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528938", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528968", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.528995", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.529022", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.529052", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.529079", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.529202", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.529230", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.529257", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.529285", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.529312", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.529340", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.529368", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.558394", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.560817", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T23:53:31.560914", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:53:35.153547", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:35.236989", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:53:39.030084", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:53:40.088905", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-22T23:53:41.285427", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-22T23:53:42.303320", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:53:43.032523", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:43.054338", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:53:43.828266", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:53:44.510150", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:53:45.225644", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:53:45.246048", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.007238", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.626061", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.635776", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.639250", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.639956", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.640019", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.640052", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.640203", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.640236", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.640268", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.640383", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.640411", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:53:46.640443", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.199024", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.199838", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.199993", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.200121", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.200238", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.200355", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.201074", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.201215", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.201597", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.201648", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.201701", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.201736", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.201767", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.201802", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.201832", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.201992", "level": "INFO", "name": "MedicalEntityRecognizer", "module": "entity_recognition", "function": "__init__", "line": 26, "message": "Initialized rule-based medical entity recognizer (spaCy and thinc disabled)", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202060", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202092", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202127", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202156", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202187", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202218", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202247", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202384", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202414", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202444", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202473", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202505", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202535", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.202566", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.232438", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.235023", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1231, Val: 264, Test: 264", "taskName": null}
{"timestamp": "2025-02-22T23:58:26.235189", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:58:29.775338", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:29.859179", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:58:33.591786", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:58:35.821680", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 17000 characters", "taskName": null}
{"timestamp": "2025-02-22T23:58:36.818149", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:58:37.529862", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:37.550601", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:58:38.304637", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:58:38.969830", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-22T23:58:39.679160", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-22T23:58:39.700443", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-22T23:58:40.477876", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-22T23:58:41.104042", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-22T23:58:41.108420", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:58:41.112713", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-22T23:58:41.113399", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:58:41.113487", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:58:41.113569", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-22T23:58:41.113712", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:58:41.113747", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:58:41.113781", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-22T23:58:41.113905", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-22T23:58:41.113936", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-22T23:58:41.113965", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}