data_generation.py

from required_libaries import time, tqdm,csv
from gpt_processing import process_gpt
from file_management import save_patient,save_records

# Placeholder for any specific imports or functions that are relevant to data generation but not directly defined in the overview

# Function definitions related to generating patient data


def generate_patients(user_request :str = None ,generate_time: int =1,record_count : int = 10,output_file_name:str ="patient_data.csv" ):
  "Generated time is the number of time the function should run.Default 1"

  if user_request is None:
    user_request =""" Heart disease
Cancer
Chronic lower respiratory disease
Stroke
Alzheimer's disease
Diabetes
Influenza and pneumonia
Kidney disease
Suicide
Septicemia
Chronic liver disease and cirrhosis
Hypertension
Parkinson's disease
Intentional self-harm
Chronic obstructive pulmonary disease (COPD)
Liver cancer
Liver disease and cirrhosis
Falls
Leukemia
Breast cancer, Country of origion :USA"""

  for i in tqdm(range(generate_time)) :
    #gen_recs , patient names and profiles generated by GPT.
    gen_recs = process_gpt(str(user_request),record_count,1)
    print("Generated time :",i+1)
    print(gen_recs)

    #Save patient
    save_patient(gen_recs,output_file_name)
    time.sleep(2)


##Read patient data
def read_patient_names():
    with open('patient_data.csv', mode='r') as csv_file:
        csv_reader = csv.reader(csv_file)
        patients = []
        next(csv_reader)  # Skip header row
        for row in csv_reader:
            patient = " ".join(row[:7])
            patients.append(patient)
    return patients

def generate_records(start_index: int = 0,max_attempts:int= 10,record_count: int = 10,output_file_name:str ="medical_data3.csv"):
  """Generate patient medica record, Input should be patient info and country of origion.
  This code run with csv file"""

  attempts = 0

  last_processed_index = start_index - 1

  while attempts < max_attempts:
      try:
          #Read patient csv file.
          patients = read_patient_names()[start_index:]
          total_patients = len(patients)


          for i, patient in tqdm(enumerate(patients, start=start_index), total=total_patients):
              print(patient)
              try:
                gpt_generated = process_gpt(str(patient),record_count, 2)
               #Save patient note .
                save_records(gpt_generated,output_file_name)
                last_processed_index = i
                print(f"patient {i+1} passed")
                time.sleep(2)
              except IndexError:
                  print(f"IndexError: Please check the CSV file and create_record() function for patient {i+1}.")
                  time.sleep(2)
                  raise  # Re-raise the error to retry the current row
              except Exception as e:
                  time.sleep(2)
                  print(f"Error: {e} occurred for patient {i}. Retrying...")
                  continue  # Retry the current row if there was an error

          break  # Exit loop if successful
      except IndexError:
          print("IndexError: Please check the CSV file and create_record() function.")
          time.sleep(2)
          attempts += 1
          if attempts == max_attempts:
              print(f"Failed after {attempts} attempts. Exiting program. Restart from {i+1}")
          else:
              start_index = last_processed_index + 1
              time.sleep(2)
              print(f"Retrying from patient {start_index}...")