ScribeMD - Transcript Parser

An AI-powered web application for analyzing medical clinic phone call transcripts. Uses advanced language models to automatically detect caller intent, extract structured information, and flag urgency levels.

Features

• AI-Powered Intent Detection: Automatically identifies the reason for the call (appointment booking, prescription refill, urgent issue, etc.)
• Structured Data Extraction: Extracts caller information including name, date of birth, phone number
• Urgency Flagging: Automatically flags urgent medical issues
• Confidence Scoring: Provides confidence level for each analysis
• Raw Transcript Input: Paste phone call transcripts into the left panel
• Parsed Data Output: View the analyzed JSON data in the right panel with all extracted information
• Submit Button: Send transcript to server for AI analysis
• Clear Button: Reset both text fields

Supported Intents

The system can detect these common medical clinic call types:

• appointment_booking
• appointment_cancellation
• appointment_reschedule
• prescription_refill
• lab_results
• billing_question
• insurance_question
• medication_side_effects
• urgent_medical_issue (flags high urgency)
• general_inquiry
• referral_request
• follow_up_call
• test_scheduling
• record_request
• other

Project Structure

scribemd/
├── app.py                 # Flask backend server
├── requirements.txt       # Python dependencies
├── templates/
│   └── index.html        # Frontend HTML
└── static/
    ├── style.css         # Styling
    └── script.js         # Frontend logic

Setup & Installation

Prerequisites

• Python 3.8+
• pip (Python package manager)
• OpenAI API key (Claude 3.5 Sonnet or GPT-4 recommended)

Steps

1. Clone the repository

   git clone <your-repo-url>
   cd scribemd

2. Create a virtual environment (optional but recommended)

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install dependencies

   pip install -r requirements.txt

4. Configure API Key

   • Open .env file and add your OpenAI API key:

   OPENAI_API_KEY=your_openai_api_key_here
   

   • Get your API key from: https://platform.openai.com/api-keys

5. Run the application

   python app.py

6. Open in browser

   • Navigate to http://localhost:5001

Usage

1. Enter or paste a medical clinic phone call transcript in the "Raw Transcript" field
2. Click the "Submit" button
3. The AI will analyze the transcript and display:
   • Detected intent (what the caller wants)
   • Extracted name, date of birth, phone number
   • Summary of the call reason
   • Urgency level (low, medium, high, critical)
   • Confidence score
   • Additional details (symptoms, medications, appointment dates, etc.)
4. Use the "Clear" button to reset both fields

Example Input:

Hi, this is Sarah Cohen, born 03/12/1988. I need to book an appointment because 
I've had chest pain for two days. Please call me back at 310-555-2211.

Example Output:

{
  "intent": "urgent_medical_issue",
  "name": "Sarah Cohen",
  "dob": "1988-03-12",
  "phone": "310-555-2211",
  "summary": "Chest pain for two days - requires urgent appointment",
  "urgency": "high",
  "confidence": 0.95,
  "extracted_details": {
    "symptoms": ["chest pain"],
    "medications_mentioned": [],
    "appointment_dates": [],
    "other_info": ["Duration: 2 days"]
  },
  "raw_transcript": "Hi, this is Sarah Cohen..."
}

API Endpoints

POST /api/parse

Sends a phone call transcript to the server for AI analysis.

Request:

{
  "transcript": "Hi, this is John Smith. I need to refill my prescription..."
}

Response:

{
  "success": true,
  "data": {
    "intent": "prescription_refill",
    "name": "John Smith",
    "dob": "1975-05-15",
    "phone": "555-0123",
    "summary": "Caller requesting prescription refill",
    "urgency": "low",
    "confidence": 0.92,
    "extracted_details": {
      "symptoms": [],
      "medications_mentioned": [],
      "appointment_dates": [],
      "other_info": []
    },
    "raw_transcript": "Hi, this is John Smith..."
  }
}

Error Response:

{
  "success": false,
  "error": "Description of what went wrong"
}

Adding Parsing Rules

Currently, the server returns the raw transcript as-is. To add parsing logic:

1. Edit app.py in the parse_transcript() function
2. Add your parsing rules and create new fields in the parsed_data dictionary
3. The changes will be reflected automatically (Flask debug mode enabled)

Adding Custom Intents

To add new intents, edit the CLINIC_INTENTS list in app.py:

CLINIC_INTENTS = [
    "appointment_booking",
    "your_new_intent_here",  # Add your custom intent
    # ... rest of intents
]

The AI will automatically learn to detect the new intent based on the name and context.

Future Enhancements

• User authentication and integration with CRM
• Interactive: fix intent correction, setting appointments, etc.
• Custom intent configuration per clinic
• Call recording analysis
• Historical data analytics
• Integration with clinic management systems
• Real-time call monitoring
• Parallelism: Batch processing for multiple transcripts
• Export to CSV/PDF reports integration with database storage
• Saas support
• Multi-language support