HealthGuard AI

AI-powered parallel health data research platform. 5 specialized agents query CDC, FDA, MedlinePlus, and Healthcare.gov simultaneously and synthesize findings into a professional report.

Architecture

User Query → FastAPI → 5 Parallel Agents → Synthesized Report → Streamed to Browser

Agent	Data Source	Port
🌍 EPHT_Agent	CDC Environmental Public Health Tracking	8889
📊 OpenData_Agent	CDC Open Data (COVID, disease surveillance)	8890
🏥 HealthcareAccess_Agent	Healthcare.gov (insurance, providers)	8891
🩺 MedlinePlus_Agent	MedlinePlus Connect (patient education, ICD-10)	8892
⚠️ OpenFDA_Agent	OpenFDA (FAERS adverse events, recalls, MAUDE)	8893

Quick Start

Option A — Docker (recommended)

cp .env.example .env        # add your ANTHROPIC_API_KEY
docker-compose up --build
open frontend.html          # open in browser, connects to localhost:8000

Option B — Local (uv)

cp .env.example .env        # add your ANTHROPIC_API_KEY
uv sync

# Terminal 1
uv run start_servers.py     # starts all 5 MCP servers (ports 8889–8893)

# Terminal 2
uv run fastapi_server.py    # API server on http://localhost:8000

# Open frontend.html in your browser

Option C — Local (pip)

cp .env.example .env
pip install -r requirements.txt
python start_servers.py     # Terminal 1
python fastapi_server.py    # Terminal 2
open frontend.html

Project Structure

├── fastapi_server.py           # Main API — agent orchestration + SSE streaming
├── start_servers.py            # Starts all 5 MCP servers
├── frontend.html               # React UI — 5 agent columns + final report
│
├── mcp_server_epht.py          # CDC EPHT MCP server
├── mcp_server_opendata.py      # CDC Open Data MCP server
├── mcp_server_healthcare.py    # Healthcare.gov MCP server
├── mcp_server_medlineplus.py   # MedlinePlus Connect MCP server
├── mcp_server_openfda.py       # OpenFDA MCP server
│
├── tools/                      # Data source API integrations
│   ├── cdc_epht.py
│   ├── cdc_open_data.py
│   ├── healthcare_gov.py
│   ├── medlineplus_connect.py
│   └── openfda_api.py
│
├── tests/                      # Integration tests
├── docs/                       # Supplementary documentation
│
├── Dockerfile
├── docker-compose.yml
├── .env.example
├── pyproject.toml
└── requirements.txt

API Endpoints

Method	Endpoint	Description
POST	/chat	Stream research (SSE) — body: { "message": "..." }
GET	/health	Health check
GET	/agents	List configured agents

Example Queries

• "What are the latest FDA recalls for blood pressure medications and what patient education is available?"
• "Analyze fentanyl-related adverse events and provide opioid safety education resources"
• "Compare air quality trends with respiratory health outcomes across states"
• "Which counties have the highest uninsured rates and how do COVID outcomes correlate?"

Adding a New Data Source

1. Create tools/your_source.py with a register_your_source_tools(mcp) function
2. Create mcp_server_your_source.py — copy any existing one as a template
3. Add an AgentConfig entry to AGENT_CONFIGS in fastapi_server.py
4. Add the server to start_servers.py and docker-compose.yml
5. Add a column for it in frontend.html

Tech Stack

• Backend: FastAPI, Server-Sent Events, asyncio
• Agents: Anthropic Claude claude-sonnet-4-6 (direct API)
• Tools: MCP protocol via fastmcp
• Frontend: React (CDN), no build step
• Package manager: uv