🏥 Intelligent Clinical Decision Support Platform powered by Azure AI

A multi-agent healthcare orchestration system that provides real-time clinical decision support for complex patient cases. Built for Microsoft Imagine Cup 2025.

• Multi-Agent AI Architecture - 5 specialized AI agents working in concert:

  • 📋 Patient Context Agent - Synthesizes patient history
  • 🔬 Labs & Reports Agent - Extracts insights from lab data
  • 📚 Research Guidelines Agent - RAG-powered evidence-based recommendations
  • ⚠️ Risk & Safety Agent - Identifies contraindications and drug interactions
  • 🧠 Orchestrator Agent - Synthesizes all inputs for clinical recommendations

• Azure-Powered Backend:

  • Azure OpenAI (GPT-4o) for clinical reasoning
  • Azure Health Data Services (FHIR R4) for patient data
  • Azure Document Intelligence for lab report extraction
  • Azure Cognitive Search for medical guidelines RAG
  • Azure Cosmos DB for persistent storage
  • Azure Monitor for audit trails

• Healthcare-Safe Design:

  • Full audit logging for compliance
  • Structured outputs with evidence citations
  • Clinician-in-the-loop decision making
  • HIPAA-ready architecture

• Frontend: React 18, TypeScript, Tailwind CSS, Radix UI, Wouter
• Backend: Express.js, TypeScript, Node.js 20
• AI/ML: Azure OpenAI, Azure Cognitive Search, Azure Document Intelligence
• Data: Azure Cosmos DB, Azure Health Data Services (FHIR R4)
• Monitoring: Azure Application Insights, Log Analytics
• Infrastructure: Bicep IaC, GitHub Actions CI/CD

• Node.js 20+
• Azure subscription with the following services:
  • Azure OpenAI (GPT-4o deployment)
  • Azure Health Data Services
  • Azure Cognitive Search
  • Azure Document Intelligence
  • Azure Cosmos DB
  • Azure Application Insights

# Install dependencies
npm install

# Copy environment template
cp .env.azure.example .env

# Configure Azure credentials in .env
# Then start development server
npm run dev

The application will be available at http://localhost:5000

# Set deployment configuration
export RESOURCE_GROUP="healthmesh-rg"
export LOCATION="eastus2"
export ENVIRONMENT="dev"
export ADMIN_EMAIL="[email protected]"

# Deploy to Azure
./script/deploy-azure.sh

1. Create Resource Group:

   az group create --name healthmesh-rg --location eastus2

2. Deploy Infrastructure:

   az deployment group create \
     --resource-group healthmesh-rg \
     --template-file infra/main.bicep \
     --parameters environment=dev [email protected]

3. Build and Deploy:

   npm run build
   az webapp deployment source config-zip \
     --resource-group healthmesh-rg \
     --name <webapp-name> \
     --src dist.zip

Copy .env.azure.example to .env and configure:

# Azure OpenAI
AZURE_OPENAI_ENDPOINT=https://your-openai.openai.azure.com/
AZURE_OPENAI_API_KEY=your-key
AZURE_OPENAI_DEPLOYMENT=gpt-4o

# Azure Cosmos DB
AZURE_COSMOS_ENDPOINT=https://your-cosmos.documents.azure.com:443/
AZURE_COSMOS_KEY=your-key
AZURE_COSMOS_DATABASE=healthmesh

# Azure Health Data Services
AZURE_FHIR_ENDPOINT=https://your-fhir.fhir.azurehealthcareapis.com

# Azure Document Intelligence
AZURE_DOCUMENT_INTELLIGENCE_ENDPOINT=https://your-docint.cognitiveservices.azure.com/
AZURE_DOCUMENT_INTELLIGENCE_KEY=your-key

# Azure Cognitive Search
AZURE_SEARCH_ENDPOINT=https://your-search.search.windows.net
AZURE_SEARCH_ADMIN_KEY=your-key

# Azure Monitor
APPLICATIONINSIGHTS_CONNECTION_STRING=InstrumentationKey=...

├── client/                 # Frontend React application
│   ├── src/
│   │   ├── components/     # UI components (Radix-based)
│   │   ├── pages/          # Route pages
│   │   ├── hooks/          # Custom React hooks
│   │   └── lib/            # Utilities
├── server/                 # Backend Express server
│   ├── azure/              # Azure service clients
│   │   ├── config.ts       # Configuration management
│   │   ├── openai-client.ts
│   │   ├── fhir-client.ts
│   │   ├── document-intelligence.ts
│   │   ├── cognitive-search.ts
│   │   ├── cosmos-db.ts
│   │   └── monitoring.ts
│   ├── azure-agents.ts     # Azure-powered AI agents
│   ├── azure-routes.ts     # REST API endpoints
│   └── index.ts            # Server entry point
├── shared/                 # Shared TypeScript schemas
├── infra/                  # Azure Bicep IaC
│   └── main.bicep          # Infrastructure definition
├── .github/workflows/      # CI/CD pipelines
│   └── deploy.yml
└── script/
    └── deploy-azure.sh     # Deployment script

• GET /api/patients - List all patients
• GET /api/patients/:id - Get patient by ID
• POST /api/patients - Create new patient
• PUT /api/patients/:id - Update patient
• DELETE /api/patients/:id - Delete patient

• GET /api/cases - List all cases
• GET /api/cases/:id - Get case by ID
• POST /api/cases - Create new case
• PUT /api/cases/:id - Update case
• POST /api/cases/:id/analyze - Run AI analysis

• GET /api/cases/:id/lab-reports - Get case lab reports
• POST /api/cases/:id/lab-reports - Upload lab report (PDF/image)

• POST /api/cases/:id/chat - Send message to clinical assistant

• GET /api/dashboard/stats - Get dashboard statistics

• GET /api/audit - Get audit logs
• GET /api/audit/:entityType/:entityId - Get entity audit trail

• All AI decisions logged to Azure Monitor
• Full audit trail for HIPAA compliance
• Managed identities for Azure service auth
• HTTPS-only with TLS 1.2+
• Role-based access control ready

This project is built for Microsoft Imagine Cup 2025 in the Healthcare category. It demonstrates:

1. Technical Excellence - Multi-agent AI architecture with Azure services
2. Healthcare Impact - Addresses clinical decision support challenges
3. Azure Integration - Deep integration with Azure AI and health services
4. Production Ready - Complete with IaC, CI/CD, and monitoring

MIT

Contributions welcome! Please read our contributing guidelines first.