from openai import OpenAI
from dotenv import load_dotenv
import openai
import json
import os
import argparse
def generate_adverse_event_report(prompt, model="gpt-4-1106-preview", max_tokens=3500, temperature=1, top_p=1, frequency_penalty=0, presence_penalty=0):
"""
Generate Adverse Event Reports for the Drug using the OpenAI API.
Args:
prompt (str): Prompt for the OpenAI API
Returns:
response (str): Response from the OpenAI API
"""
# OpenAI Client
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
# OpenAI Completion API
response = client.chat.completions.create(model=model,
messages=[
{"role": "system", "content": "Act as an expert Analyst with 20+ years of experience in Pharma and Healthcare industry. You have to generate Adverse Event Reports in properly formatted JSON"},
{"role": "user", "content": prompt}],
response_format={ "type": "json_object" },
temperature=temperature,
max_tokens=max_tokens,
top_p=top_p,
frequency_penalty=frequency_penalty,
presence_penalty=presence_penalty
)
return response.choices[0].message.content.strip()
def create_prompt(drug_name, drug_report):
"""
Create a prompt for the OpenAI API using the drug name and the drug report.
Args:
drug_name (str): Name of the drug
drug_report (str): Information about the drug
Returns:
prompt (str): Prompt for the OpenAI API
"""
return f"""Sample Adverse Event reports:
[
{{
"input": "Nicole Moore
moore123nicole@hotmail.com
32 McMurray Court, Columbia, SC 41250
9840105113, United States
Relationship to XYZ Pharma Inc.: Patient or Caregiver
Reason for contacting: Adverse Event
Message: Yes, I have been taking Mylan’s brand of Metroprolol for two years now and with no problem. I recently had my prescription refilled with the same Mylan Metoprolol and I’m having a hard time sleeping at night along with running nose. Did you possibly change something with the pill...possibly different fillers? The pharmacist at CVS didn’t have any information for me. Thank you, Nicole Moore",
"output": {{
"drug_name":"Metroprolol",
"adverse_events": ["hard time sleeping at night", "running nose"]
}}
}},
{{
"input": "Jack Ryan,
jack3rayan@gmail.com
120 Erwin RD, Canonsburg, PA 21391,
2133681441, United States
Relationship to XYZ Pharma Inc.: Patient
Reason for contacting: Defective Product
Message: I recently purchased a Wixela inhub 250/50 at my local CVS pharmacy and the inhaler appears to be defective. When I try and activate it, the yellow knob only goes down halfway. I just removed this one from the wrapper so I know it's not empty. The pharmacy wouldn't exchange it so I am contacting you to get a replacement. Thank you for your time and consideration in this matter",
"output": {{
"drug_name":"Wixela inhub 250/50",
"adverse_events": ["defective inhaler"]
}}
}},
]
Now create Adverse Event Reports in a similar way for the Drug - {drug_name}.
You have more information about the drug's use and its side effects below:
{drug_report}
Generate 15 different reports each with different side effects. Mention one or two side effects in each report at max. You have to prepare data for Entity Extraction of 2 entities: "drug_name" and "adverse_events" only.
Followng the following format for the final output:
[
{{
"input":"## Generated Report Here",
"output": {{ "drug_name":"## Name of Drug", "adverse_events": ["side effect 1", "side effect 2"] }}
}},
{{
"input":"## Generated Report Here",
"output": {{ "drug_name":"## Name of Drug", "adverse_events": ["side effect 1", "side effect 2"] }}
}},
]
"""
def create_dataset(folder_path):
"""
Create a dataset of Adverse Event Reports for the Drugs using the OpenAI Chat Completions API.
Args:
folder_path (str): Path to the folder containing the Drug Information files
Returns:
None
"""
# Create the folder if it doesn't exist
if not os.path.exists("data/entity_extraction_reports/"):
os.makedirs("data/entity_extraction_reports/")
# Iterate through the files in the folder
for filename in os.listdir(folder_path):
if filename.endswith('.txt'):
file_path = os.path.join(folder_path, filename)
# Read the contents of the file
with open(file_path, 'r') as file:
file_contents = file.read()
# Get the name of the drug from the filename
drug_name = filename.split('.')[0]
# Get the information about the drug from the file contents
drug_report = file_contents
# Create a dynamic prompt
prompt = create_prompt(drug_name, drug_report)
# Generate Adverse Event Reports for the Drug
reports = generate_adverse_event_report(prompt)
# Convert the string response to a Python Dict object
output_list = json.loads(reports)
# Save the generated data as a JSON file
with open(f"data/entity_extraction_reports/{drug_name}.txt", 'w') as text_file:
text_file.write(output_list)
if __name__ == '__main__':
# Load the .env file with the API key
load_dotenv()
# Parse the arguments
parser = argparse.ArgumentParser(description="Data Preparation Script")
parser.add_argument('--folder-path', type=str, default='data/raw_drug_info/', help="Path to the folder containing the raw Drug Information files scraped form web")
args = parser.parse_args()
# Create the folder if it doesn't exist
if not os.path.exists(args.folder_path):
os.makedirs(args.folder_path)
create_dataset(args.folder_path)
