MediChat-AI: Healthcare LLM Assistant

An AI-powered healthcare chatbot combining Llama 3.1 for generation and DistilBERT for intent classification, deployable via Streamlit/Flask or AWS infrastructure.

Demo

Features

• 🩺 Medical-Specific Responses: Context-aware healthcare answers
• 🧠 Dual-Model Pipeline:
  • Intent classification (DistilBERT)
  • Answer generation (T5/Llama-3)
• 🚀 Flexible Deployment:
  • Streamlit prototype
  • Flask REST API
  • AWS EC2 + S3 (production)

Quick Start

git clone https://github.com/yourusername/MediChat-AI.git
cd MediChat-AI
pip install -r requirements.txt

# Set API key (get from https://together.ai)
echo "TOGETHER_API_KEY=your_key_here" > .env

# Run Streamlit demo
streamlit run app.py

Project Structure

├── app.py                # Streamlit interface
├── chatbot_webapp.py     # Flask API
├── llm.py                # Together.ai LLM wrapper
├── requirements.txt      # Python dependencies
├── /models               # For local model storage
│   ├── intent_classifier_minimal
│   └── generator_minimal

Deployment Options

1. Local Development

streamlit run app.py  # Demo on http://localhost:8501
python chatbot_webapp.py  # Flask API on :5000

2. AWS Production

graph LR
    A[EC2 Instance] -->|Load| B(S3 Models)
    A -->|Serve| C(API Gateway)
    D[User] -->|HTTPS| C

Loading

Requirements:

• EC2 instance (GPU recommended)
• S3 bucket for models
• IAM roles with S3 access

⚠️ AWS configuration not included - set your own credentials

Architecture

sequenceDiagram
    User->>UI: Ask question
    UI->>Intent Classifier: Get focus area
    Intent Classifier->>Answer Engine: Structured prompt
    Answer Engine->>LLM: Augmented request
    LLM->>User: Verified response

Loading

Contributing

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit changes (git commit -m 'Add feature')
4. Push (git push origin feature/amazing-feature)
5. Open a Pull Request

License

Apache 2.0 - See LICENSE