1121 lines (1120 with data), 263.9 kB

{"timestamp": "2025-02-23T00:00:40.168154", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.168412", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.168561", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.168688", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.168807", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.168919", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.169641", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.169775", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170109", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170186", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170232", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170270", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170307", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170346", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170380", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170594", "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": "2025-02-23T00:00:40.170667", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170709", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170745", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170775", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170804", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170834", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.170863", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.171017", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.171046", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.171077", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.171109", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.171139", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.171170", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.171200", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.201739", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-23T00:00:40.204150", "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": "2025-02-23T00:00:40.204406", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:00:44.281028", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:44.385565", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:00:48.695624", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:00:48.829370", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:00:51.525568", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:00:52.429769", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:00:52.937182", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:52.953508", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:00:53.488638", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:00:53.958785", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:00:54.431179", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:00:54.445665", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:00:54.961439", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:00:55.369230", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-23T00:00:55.375088", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:00:55.379488", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:00:55.380187", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:00:55.380255", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1407, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:00:55.380294", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1407,)", "taskName": null}
{"timestamp": "2025-02-23T00:00:55.380409", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:00:55.380441", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:00:55.380470", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-23T00:00:55.380561", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:00:55.380590", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:00:55.380619", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.647885", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.648307", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.648455", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.648572", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.648691", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.648800", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.650125", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.650246", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.650553", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.650587", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.650619", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.650658", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.650690", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.650721", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.650749", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.650902", "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": "2025-02-23T00:03:39.650970", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.650999", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.651033", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.651061", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.651088", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.651116", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.651144", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.651277", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.651306", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.651337", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.651368", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.651405", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.651437", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.651470", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.681518", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-23T00:03:39.684230", "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": "2025-02-23T00:03:39.684373", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:03:43.682608", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:43.779108", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.039977", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.145310", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.182625", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.232414", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.251379", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.262204", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.290784", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.300712", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.345237", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.353724", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.549640", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.576492", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.642927", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.685853", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.724927", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.770461", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.793363", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.852595", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.863245", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.943704", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.961078", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:48.985857", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.041413", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.080281", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.095081", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.131262", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.159241", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.298941", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.361294", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.448279", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.458345", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.483915", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.496396", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.527424", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.590570", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.626709", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.677514", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.683636", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.760337", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.768506", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.784571", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.852939", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.861355", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.936463", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:49.975905", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.037392", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.047725", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.054051", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.084412", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.093524", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.179642", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.302498", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.365930", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.383113", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.394740", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.401680", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.424621", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.492197", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.560213", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.577173", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.595769", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.619903", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.657797", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.675986", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.828534", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.838766", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.852585", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.875157", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.900203", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.937566", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:50.990506", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:51.176203", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:51.187924", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:51.238811", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:51.265984", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:51.278697", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:51.321291", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:51.359175", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:51.397077", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:03:51.903888", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:51.919157", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.452109", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.485476", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.529292", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.544771", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.576496", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.586240", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.592479", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.698140", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.710457", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.747055", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.753217", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.859235", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.865488", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:52.873029", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:03:53.343507", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:03:53.357833", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:03:53.852576", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:03:53.936169", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:53.968141", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.074436", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.130124", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.138573", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.177363", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.196933", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.242702", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.247365", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.251772", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.252566", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.252624", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1407, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.252654", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1407,)", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.252789", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.252818", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.252859", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.252946", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.252973", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:03:54.253016", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.490614", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.491092", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.491255", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.491394", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.491519", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.491637", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.492348", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.492476", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.492772", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.492806", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.492839", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.492872", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.492903", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.492936", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.492967", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493137", "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": "2025-02-23T00:09:08.493209", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493241", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493277", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493307", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493338", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493370", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493401", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493548", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493581", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493611", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493642", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493672", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493702", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.493730", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.522829", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-23T00:09:08.525371", "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": "2025-02-23T00:09:08.525476", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:09:12.523436", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:12.619175", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:09:16.821387", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:09:16.946944", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:17.010099", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:17.109503", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:17.172160", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:17.376385", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:17.438523", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:17.474602", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:17.494046", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:17.729461", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:17.802150", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:17.825625", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:17.972793", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:18.115354", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:18.411184", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:18.436440", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:18.514332", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:18.696610", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:18.759074", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:18.791109", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:19.014588", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:19.544738", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:19.673383", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:19.814957", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:19.974277", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:20.010721", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:20.393661", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:20.431145", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:09:20.931598", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:20.946589", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:09:21.487815", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:09:21.589707", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:21.627333", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:21.657134", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:21.953542", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:09:22.421457", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:09:22.435647", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:09:22.933569", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.222825", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 10000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.342122", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.346884", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.351622", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.352415", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.352476", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1407, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.352508", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1407,)", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.352624", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.352654", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.352681", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.352767", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.352793", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:09:23.352821", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.492964", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.493456", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.493606", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.493733", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.493854", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.493970", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.494680", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.494825", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495131", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495168", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495203", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495236", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495267", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495301", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495332", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495505", "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": "2025-02-23T00:11:59.495577", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495606", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495641", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495671", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495699", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495728", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495758", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495904", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495932", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495961", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.495994", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.496025", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.496056", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.496085", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.526776", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-23T00:11:59.529355", "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": "2025-02-23T00:11:59.529449", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:12:03.752673", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:03.850719", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:12:08.291343", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:12:10.387839", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:12:10.719157", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:12:12.201997", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:12:12.677236", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:12.691987", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:12:13.193811", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:12:13.612643", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:12:13.876608", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:13.885531", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.169391", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.404224", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.409617", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.415171", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.416014", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.416082", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1495, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.416117", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1495,)", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.416229", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.416264", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.416300", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.416370", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.416405", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (88, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:12:14.416441", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (88,)", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.447513", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.448076", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.448244", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.448384", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.448505", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.448618", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.449203", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.449325", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.449662", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.449698", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.449730", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.449766", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.449795", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.449827", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.449858", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450019", "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": "2025-02-23T00:12:53.450080", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450109", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450138", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450165", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450193", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450222", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450251", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450381", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450409", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450439", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450469", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450501", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450530", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.450557", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.478478", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-23T00:12:53.481476", "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": "2025-02-23T00:12:53.481571", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:12:57.738959", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:12:57.838011", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:13:02.319945", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:13:04.412321", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:13:04.739986", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:13:06.210163", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:13:06.680168", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:13:06.695548", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:13:07.197891", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:13:07.615647", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:13:07.884135", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:13:07.893094", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.175577", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.412948", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.419091", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.421874", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.422582", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.422650", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1495, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.422681", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1495,)", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.422795", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.422828", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.422858", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.422920", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.422948", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (88, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:13:08.422980", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (88,)", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.340825", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.341336", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.341504", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.341633", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.341750", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.341871", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.343258", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.343416", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.343738", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.343775", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.343810", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.343843", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.343874", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.343908", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.343939", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344103", "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": "2025-02-23T00:15:07.344167", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344199", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344231", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344261", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344289", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344320", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344348", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344474", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344501", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344529", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344559", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344587", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344618", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.344646", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.374631", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.377569", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "split_data", "line": 74, "message": "Data split sizes - Train: 1319, Val: 264, Test: 176", "taskName": null}
{"timestamp": "2025-02-23T00:15:07.377664", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:15:11.124519", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:11.213737", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:15:15.197890", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:15:16.976867", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:15:18.036859", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:15:18.632210", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:15:19.378705", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:19.400189", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:15:20.190524", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:15:20.861946", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:15:21.328798", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:15:21.343622", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:15:21.838821", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:15:22.238849", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-23T00:15:22.244707", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:15:22.250147", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:15:22.250895", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:15:22.250959", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1319, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:15:22.250992", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1319,)", "taskName": null}
{"timestamp": "2025-02-23T00:15:22.251149", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:15:22.251182", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:15:22.251215", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-23T00:15:22.251317", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:15:22.251346", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (176, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:15:22.251374", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (176,)", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.836280", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.836625", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.836776", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.836910", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.837031", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.837142", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.837716", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.837840", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838146", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838179", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838212", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838244", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838274", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838307", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838335", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838505", "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": "2025-02-23T00:17:09.838564", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838592", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838621", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838649", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838681", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838713", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838741", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838871", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838899", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838927", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838958", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.838986", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.839016", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.839044", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.867119", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-23T00:17:09.869982", "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": "2025-02-23T00:17:09.870077", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:17:13.391924", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:13.475452", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:17:17.196929", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:17:18.243823", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:17:19.459417", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 15000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:17:20.530114", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:17:21.256783", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:21.278088", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:17:22.039208", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:17:22.730778", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:17:23.496071", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:17:23.516776", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.278921", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.902090", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.908291", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.911360", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.912068", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.912132", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.912162", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.912370", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.912402", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.912430", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.912541", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.912568", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:17:24.912597", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.660529", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.660873", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.661023", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.661166", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.661303", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.661422", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.661973", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662085", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662361", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662402", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662433", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662463", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662492", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662527", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662556", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662709", "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": "2025-02-23T00:20:12.662774", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662803", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662834", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662862", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662889", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662918", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.662945", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.663074", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.663101", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.663133", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.663166", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.663194", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.663222", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.663250", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.692515", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-23T00:20:12.696044", "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": "2025-02-23T00:20:12.696139", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:20:16.237546", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:16.321781", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:20:20.060175", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:20:20.160122", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:20.267212", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:20.287232", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:20.459752", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:20.484695", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:20.516319", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:20.887140", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:20.974062", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.050022", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.064150", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.152225", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.203134", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.335180", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.462970", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.547047", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.688087", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.745527", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.857692", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.909152", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.976743", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:21.986127", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.006949", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.029280", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.065764", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.149405", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.221720", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.264655", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.323282", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.337740", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.354060", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.365758", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.376237", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.385330", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.395985", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.445277", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.466667", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.710380", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:22.997853", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:23.042269", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:23.100029", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:23.184896", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:23.200802", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:20:23.920617", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:23.941250", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:20:24.696745", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:20:24.741103", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:24.793633", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:24.806061", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:24.873049", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:24.977267", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:25.057935", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:25.158530", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:25.252486", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:25.315654", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:25.331958", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:25.351502", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:20:26.077310", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:20:26.098809", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:20:26.863748", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:20:26.902627", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.082268", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.304391", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.386119", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.475401", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 7500 characters", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.488069", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.493052", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.496219", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.496808", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.496870", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.496903", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.497049", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.497080", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.497107", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.497217", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.497245", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:20:27.497272", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.630051", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.630513", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.630697", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.630849", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.631451", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.631597", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.632283", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.632430", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.632769", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.632809", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.632845", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.632879", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.632911", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.632947", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.632982", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633156", "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": "2025-02-23T00:24:00.633225", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633258", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633292", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633323", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633355", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633388", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633419", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633571", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633827", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633877", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633916", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633961", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.633997", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.634034", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.664152", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-23T00:24:00.666569", "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": "2025-02-23T00:24:00.666703", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:24:04.233730", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:04.318096", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:24:08.062525", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:24:08.164064", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:08.271028", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:08.292598", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:08.434126", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:08.493303", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:08.526168", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:08.897252", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:08.975590", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:09.076322", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:09.166065", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:09.412227", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:09.457392", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:09.563425", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:09.613486", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:09.706061", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:09.922561", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:09.990771", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:10.001662", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:10.023982", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:10.238067", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:10.342013", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:10.357896", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:10.374914", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:10.387513", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:10.408649", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:10.902552", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:11.202072", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:11.217268", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:24:11.930269", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:11.951204", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:24:12.704352", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:24:12.750844", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:12.805702", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:12.819554", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:12.885215", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:13.067255", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:13.166914", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:13.241727", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:13.256431", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:13.320720", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:13.355238", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:24:14.064858", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:24:14.085144", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:24:14.846627", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.064778", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.284394", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.468147", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.472927", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.475983", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.476696", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.476759", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.476790", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.476977", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.477008", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.477037", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.477148", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.477177", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:24:15.477206", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.059241", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.059879", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.060025", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.060166", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.060295", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.060420", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.063380", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.063544", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.063868", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.063905", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.063940", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.063972", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064010", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064042", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064071", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064227", "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": "2025-02-23T00:27:06.064288", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064318", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064349", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064379", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064410", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064440", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064473", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064602", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064631", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064662", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064692", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064721", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064752", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.064781", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.095062", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-23T00:27:06.097577", "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": "2025-02-23T00:27:06.097678", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:27:09.626563", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:09.710793", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:27:13.449088", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:27:13.546519", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:13.590933", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:13.637235", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:13.649901", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:13.801313", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:13.809638", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:13.863861", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:13.892499", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:13.902430", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:13.927571", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:13.937136", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.098141", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.147415", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.259956", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.267913", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.336428", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.370144", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.379785", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.417242", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.429681", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.513496", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.536435", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.543635", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.562253", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.694242", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.755524", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.797926", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.816477", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.899523", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:14.946673", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.034020", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.086303", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.122072", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.191316", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.218529", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.240291", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.306532", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.314463", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.333831", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.354754", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.390288", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.470129", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.502660", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.518980", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.538863", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.579027", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.610587", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.637654", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.650594", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.664824", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.674427", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.683291", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.691181", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.748566", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.768656", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.778465", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.865615", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:15.901559", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:16.007832", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:16.047112", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:16.177536", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:16.264724", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:16.290761", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:16.333991", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:16.391054", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:16.477092", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:16.492567", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:27:17.208038", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:17.229030", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:27:17.997334", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.040233", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.092808", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.103983", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.127734", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.166947", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.175160", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.263203", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.342495", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.372640", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.439397", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.513967", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.526336", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.586502", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.600867", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:18.619791", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:27:19.339523", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:27:19.359992", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.133913", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.170130", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.270561", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.344993", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.623771", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.642573", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 6000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.740650", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.746148", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.749055", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.749799", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.749859", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.749890", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.750027", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.750056", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.750083", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.750194", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.750220", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:27:20.750247", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.504192", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.504844", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.505025", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.505178", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.505307", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.505425", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.506151", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.506273", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.506582", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.506621", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.506655", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.506691", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.506721", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.506752", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.506781", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.506950", "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": "2025-02-23T00:29:31.507020", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507051", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507088", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507118", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507148", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507178", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507207", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507346", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507374", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507402", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507432", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507466", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507496", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.507525", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.537846", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-23T00:29:31.540266", "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": "2025-02-23T00:29:31.540356", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:29:35.100349", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:35.185250", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:29:38.947748", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.088538", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.133869", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.145425", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.294474", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.350931", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.377458", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.386340", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.401480", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.409750", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.418230", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.585597", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.625927", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.706707", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.738916", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.745966", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.820656", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.847116", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.856235", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.869156", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.893152", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.971083", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.987549", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:39.994504", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.009477", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.015587", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.032461", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.167652", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.228934", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.289402", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.371266", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.441150", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.502837", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.509790", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.557274", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.592398", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.598494", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.661917", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.688751", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.709400", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.718253", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.775398", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.782694", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.855921", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.889950", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.926358", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.935990", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.954806", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:40.981950", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.001107", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.040910", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.071601", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.108927", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.115763", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.130561", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.138025", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.144489", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.152733", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.181330", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.200618", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.219744", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.297503", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.364315", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.453918", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.607668", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.636758", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.655376", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.707798", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.776228", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.832782", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.919331", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:41.934874", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:29:42.658124", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:42.679543", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.440391", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.481927", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.488929", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.532022", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.542579", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.565454", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.603581", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.610838", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.698354", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.808157", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.814151", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.947847", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.958986", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:43.996971", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:44.018073", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:44.031224", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:44.050438", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:29:44.770101", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:29:44.790269", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:29:45.563076", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:29:45.701057", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:45.728620", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:45.774488", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:45.851270", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:45.962253", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:45.990027", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.057503", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.075735", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.087012", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.126189", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 5000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.173001", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.178045", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.182714", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.183361", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.183426", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.183456", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.183579", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.183611", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.183638", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.183748", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.183774", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:29:46.183804", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.963940", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.964499", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.964621", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.964732", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.964835", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.964941", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.965467", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: domain_specific_stopword_handler", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.965578", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.965858", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.965893", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.965928", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.965963", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.965994", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966026", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966057", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966208", "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": "2025-02-23T00:31:41.966288", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966322", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966358", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966403", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966442", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966480", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966511", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966643", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: whitespace_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966678", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: length_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966715", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: special_character_cleaner", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966752", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_score_processor", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966784", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_abbreviation_normalizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966820", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_term_standardizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.966851", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "add_step", "line": 233, "message": "Added preprocessing step: medical_tokenizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.994577", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_data", "line": 43, "message": "Loaded dataset with shape: (1759, 3)", "taskName": null}
{"timestamp": "2025-02-23T00:31:41.996925", "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": "2025-02-23T00:31:41.997009", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:31:45.531768", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:45.617727", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:31:49.372200", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:31:49.474997", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:49.583628", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:49.605357", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:49.749370", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:49.809047", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:49.841517", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:50.213823", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:50.292032", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:50.393657", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:50.484854", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:50.733880", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:50.778761", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:50.885595", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:50.935789", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:51.027913", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:51.243036", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:51.311102", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:51.322026", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:51.344772", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:51.560966", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:51.665963", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:51.681897", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:51.699097", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:51.711887", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:51.733272", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:52.230469", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:52.532373", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:52.547897", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:31:53.261538", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:53.282060", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:31:54.046665", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:31:54.092959", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:54.146832", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:54.160702", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:54.227435", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:54.411908", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:54.511632", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:54.587870", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:54.602933", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:54.668272", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:54.704148", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "fit", "line": 189, "message": "Starting feature extraction pipeline fitting", "taskName": null}
{"timestamp": "2025-02-23T00:31:55.432974", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "fit", "line": 126, "message": "Fitting TF-IDF vectorizer", "taskName": null}
{"timestamp": "2025-02-23T00:31:55.453865", "level": "INFO", "name": "MedicalTextFeatureExtractor", "module": "tfidf_features", "function": "transform", "line": 204, "message": "Transforming texts to features", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.213752", "level": "INFO", "name": "MedicalTfidfVectorizer", "module": "tfidf_features", "function": "transform", "line": 132, "message": "Transforming texts to TF-IDF features", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.427692", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.648318", "level": "INFO", "name": "src.preprocessing.preprocessing", "module": "preprocessing", "function": "process", "line": 261, "message": "Truncated text to 9000 characters", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.829802", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "_prepare_data_internal", "line": 150, "message": "Data preparation completed successfully", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.834831", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "save_prepared_data", "line": 190, "message": "Saved prepared data to prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.837920", "level": "INFO", "name": "DataPreparator", "module": "data_preparation", "function": "load_prepared_data", "line": 220, "message": "Loaded prepared data from prepared_data", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.838616", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTrain set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.838680", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (1231, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.838710", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (1231,)", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.838917", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nVal set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.838946", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.838973", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.839082", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 269, "message": "\\nTest set loaded successfully:", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.839111", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 270, "message": "Features shape: (264, 1017)", "taskName": null}
{"timestamp": "2025-02-23T00:31:56.839138", "level": "INFO", "name": "__main__", "module": "data_preparation", "function": "<module>", "line": 271, "message": "Labels shape: (264,)", "taskName": null}