An intelligent, agentic AI healthcare assistant powered by AG2, Groq (LLaMA 3.3 70B), FastAPI, and Next.js.
Analyze symptoms · Decode lab reports · Check drug interactions · Get voice-guided health insights — all in real time.
| 🏠 Dashboard | 💬 AI Chat Assistant |
 |
 |
| 🩺 AI Chat with Assessment Output | 🎙️ Voice Health Assistant — Idle |
 |
 |
| 🎙️ Voice Assistant — AI Response | 🧪 Lab Reports — Overview |
 |
 |
| 🧬 Lab Report — Biomarker Detail | 📊 Lab Report — AI Analysis |
 |
 |
| 🗓️ Health Timeline | 💊 Medication Check |
 |
 |
MedHive AI is a full-stack, production-grade AI healthcare copilot that combines a multi-agent orchestration framework with retrieval-augmented generation (RAG), real-time voice interaction, and a premium Next.js 15 frontend to deliver a seamless personal health management experience.
Unlike single-model chatbots, MedHive coordinates 10 specialized AI agents — each expert in a different medical domain — through a Healthcare Coordinator that routes queries, aggregates results, and returns structured, verified health assessments.
⚠️ Disclaimer: MedHive AI is for educational and informational purposes only. It is not a substitute for professional medical advice, diagnosis, or treatment. Always consult a qualified healthcare provider.
| Feature | Description |
|---|---|
| 🧠 Multi-Agent AI | 10 specialized agents coordinated by AG2 — each an expert in their domain |
| 🩺 Symptom Analysis | Triage, risk scoring, and condition matching with confidence indicators |
| 🚨 Emergency Detection | Automatic escalation for critical symptoms with immediate action guidance |
| 🧬 Lab Report AI | Upload PDF/image lab reports and get plain-language biomarker explanations |
| 💊 Drug Interaction Check | Identifies dangerous medication combinations from your health profile |
| 🔍 Verified Medical Search | Real-time web search grounded in trusted medical sources |
| 📚 RAG Knowledge Base | ChromaDB vector store with curated medical literature |
| 🎙️ Voice Health Assistant | Whisper STT + Groq TTS — speak your symptoms, hear your analysis |
| 👤 Patient Memory | SQLite-backed patient profiles with full health history |
| 📊 Health Timeline | Visual longitudinal tracking of symptoms, vitals, and reports |
| 📱 Responsive UI | Premium Next.js 15 interface with Framer Motion animations |
| 🔒 Privacy First | All data stays local — no cloud storage without explicit consent |
┌─────────────────────────────────────────────────────────────┐
│ Next.js 15 Frontend │
│ Dashboard · Chat · Voice · Lab Reports · Medications · │
│ Health Timeline · Patients · Settings │
└─────────────────────┬───────────────────────────────────────┘
│ HTTP / REST API
┌─────────────────────▼───────────────────────────────────────┐
│ FastAPI Backend (Python) │
│ │
│ ┌─────────────┐ ┌───────────────┐ ┌───────────────────┐ │
│ │ /analyze │ │/analyze-report│ │/voice/voice-assist│ │
│ │ (Chat AI) │ │ (Lab PDF) │ │ (Whisper+TTS) │ │
│ └──────┬──────┘ └──────┬────────┘ └─────────┬─────────┘ │
│ │ │ │ │
│ ┌──────▼────────────────▼─────────────────────▼─────────┐ │
│ │ Healthcare Coordinator (AG2) │ │
│ │ Routes · Orchestrates · Aggregates · Verifies │ │
│ └───────────────────────────────────────────────────────┘ │
│ │ │
│ ▼ │
│ ┌───────────────────────────────────────────────────────┐ │
│ │ Agent Pool │ │
│ │ 🧠 Medical 🚨 Emergency 📊 Risk Assessment │ │
│ │ 🔍 Web Search 💊 Drug Inter. 🧬 Lab Report │ │
│ │ ✅ Verification 📝 Summary 💪 Coach 🩺Symptom │ │
│ └───────────────────────────────────────────────────────┘ │
│ │ │
│ ▼ │
│ ┌───────────────────────────────────────────────────────┐ │
│ │ Services & Storage │ │
│ │ ChromaDB (RAG) · SQLite (Patients) · Groq API │ │
│ └───────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────┘
| Agent | Role |
|---|---|
| MedicalAgent | Primary symptom triage and medical knowledge base queries |
| EmergencyAgent | Detects life-threatening conditions, triggers emergency protocol |
| RiskAgent | Calculates composite risk scores (Low / Medium / High / Critical) |
| SymptomAgent | Extracts, classifies, and tracks symptom patterns |
| WebSearchAgent | Fetches real-time data from trusted medical sources (Mayo Clinic, NIH, etc.) |
| LabReportAgent | Parses and explains lab report biomarkers in plain language |
| DrugInteractionAgent | Analyzes medication combinations for dangerous interactions |
| VerificationAgent | Cross-validates AI output against evidence sources |
| SummaryAgent | Synthesizes multi-agent findings into a structured health assessment |
| CoachAgent | Provides actionable wellness recommendations and lifestyle guidance |
| Technology | Purpose |
|---|---|
| FastAPI | High-performance async REST API |
| AG2 (AutoGen v2) | Multi-agent orchestration framework |
| Groq API | Ultra-fast LLaMA 3.3 70B inference |
| ChromaDB | Vector database for RAG |
| Whisper | Speech-to-text transcription |
| SQLite + SQLAlchemy | Patient data persistence |
| LangChain | Document loading and text splitting |
| Uvicorn | ASGI server |
| Technology | Purpose |
|---|---|
| Next.js 15 | React framework with App Router |
| TypeScript | Type-safe development |
| Tailwind CSS v4 | Utility-first styling |
| Framer Motion | Fluid animations and transitions |
| Material Symbols | Google's icon system |
| Inter | Premium typography |
medhive-ai/
├── backend/
│ ├── agents/ # 10 specialized AI agents
│ │ ├── medical_agent.py
│ │ ├── emergency_agent.py
│ │ ├── risk_agent.py
│ │ ├── symptom_agent.py
│ │ ├── web_search_agent.py
│ │ ├── lab_report_agent.py
│ │ ├── drug_interaction_agent.py
│ │ ├── verification_agent.py
│ │ ├── summary_agent.py
│ │ └── coach_agent.py
│ ├── workflows/
│ │ └── healthcare_coordinator.py # Agent orchestration
│ ├── api/
│ │ ├── routes.py # /analyze endpoint
│ │ ├── report_routes.py # /analyze-report
│ │ └── patient_routes.py # /patients CRUD
│ ├── voice/
│ │ ├── voice_routes.py # /voice/voice-assistant
│ │ ├── stt.py # Whisper STT
│ │ ├── tts.py # Text-to-Speech
│ │ └── whisperflow_service.py
│ ├── rag/
│ │ ├── ingestion.py # Document ingestion
│ │ └── retriever.py # Vector search
│ ├── services/
│ │ ├── lab_report_service.py
│ │ └── report_parser.py
│ ├── models/ # SQLAlchemy ORM models
│ ├── schemas/ # Pydantic request/response schemas
│ ├── repositories/ # Database access layer
│ ├── config/settings.py # App configuration
│ ├── db/init_db.py # Database initialization
│ ├── data/medical_docs/ # Medical knowledge base
│ ├── chroma_db/ # ChromaDB vector store
│ ├── outputs/ # Generated TTS audio files
│ ├── uploads/ # Uploaded lab reports
│ ├── main.py # FastAPI app entry point
│ └── .env.example
│
├── frontend/
│ └── src/
│ ├── app/
│ │ ├── dashboard/ # Health dashboard
│ │ ├── chat/ # AI chat interface
│ │ ├── voice/ # Voice assistant
│ │ ├── lab-reports/ # Lab report analyzer
│ │ ├── medications/ # Drug interaction checker
│ │ ├── health-timeline/ # Health history
│ │ ├── patients/ # Patient management
│ │ └── settings/ # App settings
│ └── components/
│ ├── Sidebar.tsx # Navigation sidebar
│ └── ui/ # Reusable UI components
│
├── img/ # Screenshots
└── README.md
- Python 3.11+
- Node.js 18+
- A free Groq API key
git clone https://github.com/Dakshin10/medhive-ai.git
cd medhive-aicd backend
# Create and activate virtual environment
python -m venv .venv
.venv\Scripts\activate # Windows
# source .venv/bin/activate # macOS/Linux
# Install dependencies
pip install -r requirements.txt
# Configure environment
copy .env.example .env # Windows
# cp .env.example .env # macOS/Linux
# Edit .env and add your Groq API key
# GROQ_API_KEY=your_key_here
# MODEL_NAME=llama-3.3-70b-versatile
# Initialize the database
python -c "from db.init_db import init_db; init_db()"
# Start the backend server
uvicorn main:app --reloadThe API will be running at http://localhost:8000
Interactive API docs: http://localhost:8000/docs
cd frontend
# Install dependencies
npm install
# Start the development server
npm run devThe frontend will be running at http://localhost:3000
Create backend/.env from the example:
# Required
GROQ_API_KEY=your_groq_api_key_here
# Model Selection (default: llama-3.3-70b-versatile)
MODEL_NAME=llama-3.3-70b-versatileGet your free Groq API key at console.groq.com — inference is blazing fast (700+ tokens/sec) and free to start.
| Method | Endpoint | Description |
|---|---|---|
POST |
/analyze |
Main health query — routes through all agents |
POST |
/analyze-report |
Upload & analyze lab report PDF |
POST |
/voice/voice-assistant |
Voice query: audio in → transcript + AI response + audio out |
GET |
/patients |
List all patients |
POST |
/patients |
Create patient profile |
GET |
/patients/{id} |
Get patient details |
PUT |
/patients/{id} |
Update patient record |
DELETE |
/patients/{id} |
Delete patient |
GET |
/health |
API health check |
GET |
/docs |
Interactive Swagger UI |
curl -X POST http://localhost:8000/analyze \
-H "Content-Type: application/json" \
-d '{"message": "I have been having chest tightness, shortness of breath, and fatigue for 3 days"}'Response:
{
"status": "SUCCESS",
"assessment": {
"symptoms": ["chest tightness", "shortness of breath", "fatigue"],
"risk_level": "HIGH",
"possible_conditions": ["Cardiac event", "Pulmonary embolism", "Severe anemia"],
"recommendations": ["Seek immediate medical attention", "Avoid strenuous activity"],
"confidence_score": 87,
"evidence_sources": ["https://www.mayoclinic.org/..."]
}
}curl -X POST http://localhost:8000/analyze-report \
-F "file=@CBC_report.pdf"User speaks → Whisper STT → Text transcript
↓
Healthcare Coordinator (10 agents)
↓
Structured JSON assessment
↓
Groq TTS → Audio response file
↓
Frontend plays audio + displays rich assessment card
- Cardiology — Heart rate, blood pressure, cardiac symptoms
- Endocrinology — Diabetes (HbA1c, glucose), thyroid function
- Hematology — CBC, hemoglobin, platelets, anemia indicators
- Lipidology — LDL, HDL, triglycerides, cardiovascular risk
- Hepatology — Liver enzymes (ALT, AST), metabolic panel
- Nephrology — Kidney function (creatinine, BUN, GFR)
- Pulmonology — Respiratory symptoms, oxygen saturation
- Pharmacology — Drug-drug interactions, supplement safety
- Emergency Medicine — Triage and critical condition detection
- Preventive Medicine — Wellness coaching and lifestyle optimization
- ✅ All data processed locally — nothing sent to external servers except LLM inference
- ✅ No patient data stored in the cloud
- ✅
.envsecrets are never committed to version control - ✅ CORS configured for local development
- ✅ SQLite database stays on your machine
⚠️ For production use, add authentication, HTTPS, and proper secrets management
- 🔐 User authentication (JWT / OAuth2)
- 🌍 Multi-language support (Hindi, Spanish, French)
- 📱 React Native mobile app
- 🩺 FHIR / HL7 integration for EHR import
- 📈 Advanced analytics dashboard with charts
- 🔔 Push notification for medication reminders
- 🤝 Doctor-patient secure messaging
- 🧬 Genomics data integration
- ☁️ Optional cloud sync with end-to-end encryption
Contributions are welcome! Here's how to get started:
# Fork the repo and clone it
git clone https://github.com/your-username/medhive-ai.git
# Create a feature branch
git checkout -b feature/your-amazing-feature
# Make your changes, then commit
git commit -m "feat: add amazing feature"
# Push and open a Pull Request
git push origin feature/your-amazing-featurePlease follow Conventional Commits and ensure all new agents include proper error handling.
This project is licensed under the MIT License — see the LICENSE file for details.
| Tool | Purpose |
|---|---|
| AG2 (AutoGen) | Multi-agent orchestration |
| Groq | Ultra-fast LLM inference |
| Meta LLaMA 3.3 | Foundation language model |
| OpenAI Whisper | Speech recognition |
| ChromaDB | Vector database |
| FastAPI | Python web framework |
| Next.js | React meta-framework |
| Framer Motion | UI animations |
