An AI-powered healthcare platform leveraging multi-agent intelligence and Gemini 2.5 (Flash + Vision) to provide intelligent diagnostics, radiology insights, treatment suggestions, and real-time interaction — built for patients, doctors, and healthcare innovators.
📦 GitHub Repository 🔗 Live Site
MediNova transforms modern healthcare diagnostics with the power of advanced AI agents and multimodal analysis. Whether you're a concerned patient or a medical professional, MediNova simplifies symptom decoding, test/treatment guidance, radiology interpretation, and medical communication — securely and intelligently.
- Smart symptom-to-disease mapping using Gemini 2.5 Flash
- Text or voice input support for ease of use
- Generates probable causes with actionable recommendations
- Suggests relevant diagnostic tests based on symptom profiles
- Helps avoid redundant or unnecessary testing
- Uses AI reasoning to provide evidence-based treatment plans
- Designed to aid patients and assist physicians in decision-making
- Upload medical scans (X-ray, MRI, etc.) for AI-powered annotation
- Gemini Vision identifies visible signs and anomalies
- Upload prescriptions (including handwritten ones)
- Extracts medications, dosages, instructions
- Highlights affordable alternatives and pricing insights
- Auto-generated patient summary and analysis in PDF format
- Includes annotated image, extracted insights, and reports
- Vision-powered deep scan interpretation
- Ideal for remote diagnostics or second opinions
- Talk to MediNova using your voice — available in assistant + symptom modules
- Built using Web Speech API and optimized for accessibility
- Manage patient analysis history
- Track cases, review prescriptions, and collaborate securely
- Role-based access for patients and doctors
- Firestore & Firebase Auth used with data encryption practices
- Natural, real-time chat with a friendly medical AI assistant
MediNova runs on micro-agent AI architecture, with each module responsible for a specialized healthcare function:
- Input Parser (Text, Image, or Voice)
- Symptom Agent
- Test & Treatment Agent
- Radiology Vision Agent
- Prescription OCR + NLP
- PDF Generator
- Chat & Follow-up Agent
Each agent uses Gemini 2.5 (Flash or Vision) in a coordinated pipeline, providing human-level insights in real time.
| Layer | Technology |
|---|---|
| 🧑🎨 Frontend | React.js, Tailwind CSS, Vite |
| 🧠 AI Backend | Gemini 2.5 Flash + Vision APIs (Google) |
| 🔐 Auth & DB | Firebase Authentication, Firestore DB |
| 🗣️ Voice Input | Web Speech API |
| 🧩 UI Libraries | Framer Motion, Shadcn/UI, Lucide Icons |
| ☁️ Deployment | Vercel |
 |
 |
 |
 |
 |
 |
 |
 |
 |
git clone https://github.com/Thorfinn05/MediNova-AI.git
cd medinovanpm installVITE_GEMINI_API_KEY=your_api_key_here
VITE_FIREBASE_API_KEY = your_firebase_api_key
VITE_FIREBASE_AUTH_DOMAIN = your_project_id.firebaseapp.com
VITE_FIREBASE_PROJECT_ID = your_project_id
VITE_FIREBASE_STORAGE_BUCKET = your_project_id.appspot.com
VITE_FIREBASE_MESSAGING_SENDER_ID = your_messaging_sender_id
VITE_FIREBASE_APP_ID = your_app_id
VITE_FIREBASE_MEASUREMENT_ID = your_measurement_idnpm run devApp will run at: http://localhost:3000
medinova/
├── public/ # Static files
├── src/
│ ├── components/ # Reusable UI Components
│ ├── pages/ # Route-based pages
│ ├── assets/ # Images & icons
│ ├── utils/ # AI service handlers
│ └── App.jsx # Main App entry
├── .env # Environment config
├── package.json # Dependencies
└── vite.config.js # Build config| Feature | Status |
|---|---|
| 📍 Location-based pharmacy suggestions from prescription | 🚧 Planned |
| 📱 Fully responsive mobile PWA version | 🚧 Planned |
| 🧑⚕️ Enhanced Doctor–Pharmacy collaboration portal | 🚧 Planned |
| 🧬 Wearable health integration | 🚧 Planned |
| 🌐 Multilingual Support | 🚧 Planned |
| 🧠 More accurate AI-powered radiology insight pipeline | 🚧 Planned |
- Integrating multi-modal AI agents (text + vision) into one flow
- Ensuring OCR works well on real-world, low-quality prescription images
- Making chatbot answers both concise and medically relevant
- Handling simultaneous PDF generation + image processing
- Voice input integration using Web Speech API on multiple browsers
- Designing for both doctors and patients without complexity
- Rudranil Das – @Thorfinn05
- Aitijhya Roy – @AitijhyaCoded
This project is licensed under the MIT License.
We welcome contributions from the open-source community. Check out the CONTRIBUTING.md.
# Fork → Clone → Branch → Code → Pull Request 🚀🧬 Empowering diagnostics with AI — one pixel, one symptom at a time.