A full-stack application for conducting virtual clinical trials using AI agents to validate medical research hypotheses.

Clinova/
├── frontend/          # React + Vite web application
├── backend/           # FastAPI backend + Python AI agents
├── visualization/     # 3D WebGL visualization (Next.js)
├── docker-compose.yml # Container orchestration
└── README.md          # This file

• Framework: React 18 + TypeScript
• Build Tool: Vite 5
• UI Components: shadcn/ui + Radix UI
• Styling: Tailwind CSS
• State Management: TanStack Query
• 3D Graphics: Three.js + React Three Fiber
• Backend Integration: REST API + Supabase
• Routing: React Router v6
• Form Handling: React Hook Form + Zod validation

• API Framework: FastAPI
• Language: Python 3.11+
• AI Models: OpenAI GPT-4, Google Gemini
• Key Features:
  • RESTful API for trial management
  • Multi-agent pipeline for clinical trial simulation
  • Clinical trial replication (PREDICT, Barrett, CARE, VALOR, NEPHRIC)
  • Contrast-induced nephropathy (CIN) analysis
  • OMOP CDM data lookup
  • Automated trial validation

• Framework: Next.js (React)
• 3D Features: React Three Fiber + Yoga flexbox layouts in WebGL
• Purpose: Advanced 3D data visualization

1. Clone the repository

git clone <repository-url>
cd Clinova

2. Set up environment variables

# Copy example env file
cp .env.example .env

# Edit .env and add your API keys
# OPENAI_API_KEY=your_openai_key
# GEMINI_API_KEY=your_gemini_key

3. Start all services

docker-compose up

This will start:

• Backend API at http://localhost:8000
• Frontend at http://localhost:5173

cd backend

# Create virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Configure environment variables
cp .env.example .env
# Edit .env and add your API keys

# Start the server
python server.py
# Or: uvicorn server:app --reload --host 0.0.0.0 --port 8000

Backend API will be available at http://localhost:8000

cd frontend

# Install dependencies
npm install

# Configure environment variables
cp .env.example .env
# Edit .env if needed

# Start development server
npm run dev

Frontend will be available at http://localhost:5173

# Backend API Keys
OPENAI_API_KEY=your_openai_api_key_here
GEMINI_API_KEY=your_gemini_api_key_here

# Frontend
VITE_SUPABASE_URL=your_supabase_url
VITE_SUPABASE_PROJECT_ID=your_supabase_project_id
VITE_SUPABASE_PUBLISHABLE_KEY=your_supabase_key

OPENAI_API_KEY=your_openai_api_key_here
GEMINI_API_KEY=your_gemini_api_key_here

VITE_API_URL=http://localhost:8000
VITE_SUPABASE_URL=your_supabase_url
VITE_SUPABASE_PROJECT_ID=your_project_id
VITE_SUPABASE_PUBLISHABLE_KEY=your_supabase_key

Once the backend is running, visit:

• Swagger UI: http://localhost:8000/docs
• ReDoc: http://localhost:8000/redoc

POST   /api/trials              # Create new trial
GET    /api/trials              # List all trials
GET    /api/trials/{id}/status  # Get trial status
GET    /api/trials/{id}/results # Get trial results
DELETE /api/trials/{id}         # Delete trial

1. Start the backend and frontend (using docker-compose or manually)

2. Create a trial via API:

curl -X POST http://localhost:8000/api/trials \
  -H "Content-Type: application/json" \
  -d '{
    "question": "How do iodinated contrast agents affect eGFR subgroups and risk of AKI in the next 30 days?"
  }'

3. Check trial status:

curl http://localhost:8000/api/trials/{run_id}/status

4. Get results (once completed):

curl http://localhost:8000/api/trials/{run_id}/results

Navigate to http://localhost:5173 and use the web interface to:

• Submit clinical trial questions
• Monitor trial progress
• View results and generated code
• Explore OMOP concept mappings

The backend runs a sophisticated multi-agent pipeline:

1. Question Agent (GPT-4): Transforms natural language to causal PICO format
2. Design Agent (GPT-4): Generates Target Trial Emulation (TTE) protocol
3. Validator Agent (GPT-4): Validates design against clinical rubric (up to 3 iterations)
4. OMOP Lookup Agent: Resolves medical terms to standardized concept IDs
5. Code Agent (Gemini): Generates Python implementation code

The system can replicate 5 major randomized controlled trials:

1. PREDICT (2008): CT with IV contrast in CKD + diabetes patients
2. Barrett et al. (2006): CE-MDCT in CKD patients
3. CARE (2007): Cardiac angiography/PCI in CKD ± diabetes
4. VALOR (2008): Coronary angiography in CKD
5. NEPHRIC (2003): High-risk diabetes + CKD angiography

cd frontend
npm run dev          # Start dev server
npm run build        # Production build
npm run test         # Run tests
npm run lint         # Lint code

cd backend
python -m agent --question "your question here"  # Run CLI
python server.py     # Run API server

# Health check
curl http://localhost:8000/

# Create a trial
curl -X POST http://localhost:8000/api/trials \
  -H "Content-Type: application/json" \
  -d '{"question": "How does contrast affect kidney function?"}'

• Virtual Clinical Trials: Simulate and validate clinical trials using AI
• Multi-Agent System: Specialized agents for different pipeline stages
• Interactive 3D Visualization: Advanced WebGL-based data visualization
• Real-time Trial Monitoring: Track trial progress through web interface
• OMOP CDM Integration: Standardized medical terminology lookup
• Agent-Based Validation: AI agents validate trial hypotheses autonomously
• Responsive UI: Modern, accessible interface built with shadcn/ui
• RESTful API: Complete backend API with OpenAPI documentation

# Build production images
docker-compose -f docker-compose.prod.yml build

# Start production services
docker-compose -f docker-compose.prod.yml up -d

For production, make sure to:

1. Use strong, unique API keys
2. Set appropriate CORS origins in backend
3. Use production Supabase instance
4. Configure proper logging
5. Set up monitoring and alerting

• Check that OPENAI_API_KEY and GEMINI_API_KEY are set
• Ensure Python 3.11+ is installed
• Verify all dependencies: pip install -r requirements.txt

• Ensure backend is running on port 8000
• Check VITE_API_URL in frontend/.env
• Verify CORS settings in backend/server.py

• Ensure Docker and Docker Compose are installed
• Check .env file exists and has correct values
• Try: docker-compose down -v && docker-compose up --build

See LICENSE file.

This is a research project focused on virtual clinical trial validation using AI agents. Contributions are welcome!

Built for advancing medical research through AI-powered virtual clinical trials.