# Medical Report Analysis Platform

A comprehensive AI-powered platform for analyzing medical PDF reports using 50+ specialized medical models across 9 clinical domains.

## Features

### Two-Layer AI Architecture
- **Layer 1**: PDF extraction, document classification, and intelligent routing
- **Layer 2**: Specialized model analysis with concurrent processing and result synthesis

### 50+ Specialized Medical Models
- **Clinical Notes**: MedGemma 27B, Bio_ClinicalBERT
- **Radiology**: MedGemma 4B Multimodal, MONAI
- **Pathology**: Path Foundation, UNI2-h
- **Cardiology**: HuBERT-ECG
- **Laboratory**: DrLlama, Lab-AI
- **Drug Interactions**: CatBoost DDI
- **Diagnosis & Triage**: MedGemma 27B
- **Medical Coding**: Rayyan Med Coding
- **Mental Health**: MentalBERT

### Comprehensive Analysis
- Multi-modal content extraction (text, images, tables)
- Document type classification
- Specialized model routing
- Concurrent processing
- Result synthesis and validation
- Clinical insights generation

### Regulatory Compliance
- HIPAA compliant architecture
- GDPR aligned data processing
- FDA guidance adherence
- Medical-grade security

## Architecture

```
┌─────────────────────────────────────────────────────────────┐
│                    Frontend (React + TypeScript)            │
│  - Professional medical-grade UI                            │
│  - Real-time analysis visualization                         │
│  - Comprehensive results display                            │
└─────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────┐
│                    Backend (FastAPI + Python)               │
│                                                             │
│  ┌─────────────────────────────────────────────────────┐  │
│  │  Layer 1: PDF Understanding & Classification        │  │
│  │  - PDF Processor (PyMuPDF, OCR)                    │  │
│  │  - Document Classifier                             │  │
│  │  - Intelligent Routing                             │  │
│  └─────────────────────────────────────────────────────┘  │
│                              │                              │
│                              ▼                              │
│  ┌─────────────────────────────────────────────────────┐  │
│  │  Layer 2: Specialized Medical Analysis             │  │
│  │  - Model Router (50+ models)                       │  │
│  │  - Concurrent Processing                           │  │
│  │  - Analysis Synthesizer                            │  │
│  └─────────────────────────────────────────────────────┘  │
└─────────────────────────────────────────────────────────────┘
```

## Project Structure

```
medical-ai-platform/
├── backend/
│   ├── main.py                    # FastAPI application
│   ├── pdf_processor.py           # PDF extraction
│   ├── document_classifier.py     # Document classification
│   ├── model_router.py            # Model routing & execution
│   ├── analysis_synthesizer.py    # Result synthesis
│   └── requirements.txt           # Python dependencies
│
├── medical-ai-frontend/
│   ├── src/
│   │   ├── App.tsx               # Main application
│   │   ├── components/
│   │   │   ├── Header.tsx        # Header component
│   │   │   ├── FileUpload.tsx    # File upload interface
│   │   │   ├── AnalysisStatus.tsx # Progress visualization
│   │   │   ├── AnalysisResults.tsx # Results display
│   │   │   └── ModelInfo.tsx     # Model information
│   │   └── ...
│   └── ...
│
└── docs/                          # Comprehensive documentation
    ├── architecture_design/
    ├── pipeline_design/
    ├── specialized_models_research/
    └── compliance_research/
```

## Quick Start

### Backend Setup

```bash
cd backend

# Install dependencies
pip install -r requirements.txt

# Run the server
python main.py
```

The backend will be available at `http://localhost:7860`

### Frontend Setup

```bash
cd medical-ai-frontend

# Install dependencies
pnpm install

# Run development server
pnpm dev
```

The frontend will be available at `http://localhost:5173`

## API Endpoints

### Health Check
```
GET /health
```

### Analyze Document
```
POST /analyze
Content-Type: multipart/form-data

Body:
- file: PDF file

Response:
{
  "job_id": "uuid",
  "status": "processing",
  "progress": 0.0,
  "message": "Analysis started..."
}
```

### Check Status
```
GET /status/{job_id}

Response:
{
  "job_id": "uuid",
  "status": "completed",
  "progress": 1.0,
  "message": "Analysis complete"
}
```

### Get Results
```
GET /results/{job_id}

Response:
{
  "job_id": "uuid",
  "document_type": "radiology",
  "confidence": 0.95,
  "analysis": {...},
  "specialized_results": [...],
  "summary": "...",
  "timestamp": "2025-10-28T18:38:23Z"
}
```

### Supported Models
```
GET /supported-models

Response:
{
  "domains": {
    "clinical_notes": {...},
    "radiology": {...},
    ...
  }
}
```

## Deployment

### Hugging Face Spaces

This platform is designed for deployment on Hugging Face Spaces with GPU support.

1. Create a new Space on Hugging Face
2. Select "Docker" as the SDK
3. Choose GPU hardware (T4 or A100 recommended)
4. Upload the project files
5. Configure environment variables (HF_TOKEN if needed)

### Environment Variables

- `HF_TOKEN`: Hugging Face API token for model access
- `VITE_API_URL`: Backend API URL (for frontend)

## Development

### Adding New Models

To add a new specialized model:

1. Update `model_router.py` with model configuration
2. Implement model execution logic
3. Update documentation

### Extending Analysis

To extend analysis capabilities:

1. Modify `analysis_synthesizer.py` for new fusion strategies
2. Update result schema as needed
3. Enhance frontend visualization

## Security & Compliance

### HIPAA Compliance
- Encrypted data transmission
- Secure temporary file handling
- Audit logging
- Access controls

### GDPR Alignment
- Data minimization
- Privacy by design
- User consent mechanisms
- Right to erasure

### FDA Guidance
- Transparency in AI decision-making
- Bias detection and mitigation
- Clinical validation frameworks
- Performance monitoring

## Performance

- **Layer 1 Processing**: < 2 seconds per page
- **Document Classification**: < 500 ms
- **Specialized Analysis**: 2-10 seconds (depending on complexity)
- **Total Analysis Time**: 30-60 seconds for typical reports

## Limitations & Disclaimer

**IMPORTANT**: This platform provides AI-assisted analysis and is designed for clinical decision support. All results must be reviewed and verified by qualified healthcare professionals.

- Not a substitute for professional medical judgment
- Requires specialist review for clinical decisions
- Performance varies by document quality and type
- Continuous validation required for clinical deployment

## Support & Documentation

For comprehensive documentation, see the `docs/` directory:

- Architecture Design
- Pipeline Design
- Model Mapping
- Compliance Guidelines

## License

This project is intended for research and development purposes. Clinical deployment requires appropriate regulatory clearances and compliance verification.

## Contributors

Built with comprehensive research and design following FDA guidance, HIPAA requirements, GDPR principles, and medical AI best practices.

---

**Medical Report Analysis Platform** - Advanced AI-Powered Clinical Intelligence