Built for Google Solution Challenge 2026 | Theme: Unbiased AI Decision-Making Team: Orion-XAI | Soaeb Hasan
OrganBridge is a full-stack web application that uses machine learning to intelligently match organ donors with recipients — while auditing its own decisions for bias. Built with Django and powered by a TF-IDF + Cosine Similarity algorithm, it goes far beyond traditional blood-type-only matching by analyzing 15+ compatibility factors in real time.
At its core, OrganBridge doesn't just predict matches — it explains them using Explainable AI (XAI) and ensures fairness across age, gender, and geography using bias detection.
In a system where every second counts and every decision impacts a life, OrganBridge makes organ allocation smarter, fairer, and more transparent.
- AI-Powered Matching — TF-IDF vectorization + Cosine Similarity engine evaluating 15+ parameters (blood type, age, geography, lifestyle, medical history)
- Role-Based Dashboards — Separate dashboards for donors and recipients
- Custom User Model — Extended Django user with medical fields (blood type, age, gender, race, city)
- Authentication System — Register, Login, Logout with session management
- Responsive UI — Tailwind CSS + HTMX powered interface
- Donor & Recipient Profiles — Complete medical profile setup
- Match Engine Integration — Full ML pipeline connected to Django views
- Real-Time Messaging — HTMX-powered live chat between matched donors and recipients
The matching engine uses two layers of logic:
-
ML Layer — Donor and recipient profiles are serialized into feature strings and transformed using a trained TF-IDF vectorizer. Cosine similarity is computed to produce a raw compatibility score (0–100%).
-
Business Rules Layer — Blood type compatibility, urgency level, geographic proximity, and health status are applied on top of the ML score to produce a final ranked list.
-
Fallback — If trained model files are unavailable, the engine automatically falls back to a rule-based scoring system.
| Category | Factors |
|---|---|
| Medical | Blood type, organ type, medical history, BMI |
| Personal | Age, gender, race |
| Lifestyle | Smoking status, alcohol use, drug history, avg sleep |
| Geographic | City, state, travel distance |
| Clinical | Urgency level, health status, last medical checkup |
This section outlines the planned extension of OrganBridge to involve hospitals as verified intermediaries in the organ matching process.
Donor → OrganBridge → Recipient
Donor → Hospital (Verified) → OrganBridge AI → Hospital → Recipient
Hospital Dashboard Hospitals will have their own verified accounts to manage patients, confirm donor eligibility, track transplant history, and receive real-time match notifications.
Medical Verification System Every donor and recipient profile will require hospital verification before entering the matching pool — eliminating fake data and increasing match reliability. Each verified profile will carry a Trust Score.
Transplant Coordination Once a match is accepted, hospitals on both sides will coordinate surgery scheduling, operation theater availability, and inter-hospital organ transfer logistics.
Cold Chain Tracking Organs have a limited viability window (kidneys: 24-36 hours). Phase 2 will introduce real-time organ tracking from harvest to transplant using Google Maps API — ensuring organs reach recipients before expiry.
Chain Transplant Support OrganBridge will support paired and chain transplants — where a donor who cannot directly help their loved one triggers a chain of compatible swaps across multiple hospitals, saving multiple lives from a single donation.
Doctor & Surgeon Portal Doctors will be able to review AI-generated match explanations, add medical notes, request second opinions, and use Gemini AI for surgical risk assessment before approving a transplant.
Regulatory & Government Integration Phase 2 will integrate with NOTTO (National Organ & Tissue Transplant Organisation) for legal compliance, automated documentation, and government audit trails for every transplant decision.
| Phase | Timeline | Milestone |
|---|---|---|
| Phase 1 | Now ✅ | Donor-Recipient AI Matching + Gemini + Bias Dashboard |
| Phase 2 | 2 months | Hospital Integration + Medical Verification |
| Phase 3 | 3 months | NOTTO / Government Integration |
| Phase 4 | 5 months | Pan-India Deployment |
organBridge/
├── accounts/ # Custom user model, auth views
├── profiles/ # DonorProfile, RecipientProfile, dashboards
├── matches/ # OrganMatch, MatchMessage, matching views
├── ml_model/ # ML engine, training, bias audit (in-app)
│ ├── matching_algorithm.py
│ ├── train_model.py
│ └── trained_models/
├── bias_dashboard/ # Fairness metrics visualization
├── organBridge/ # Django settings, URLs
├── templates/ # All HTML templates
├── static/ # CSS, JS
└── requirements.txt
| Layer | Technology |
|---|---|
| Backend | Django 6.0, Python 3.14 |
| ML / AI | Scikit-learn (TF-IDF, Cosine Similarity) |
| XAI | SHAP |
| Bias Detection | Fairlearn |
| Frontend | Tailwind CSS, HTMX, Alpine.js |
| Database | SQLite (dev) → PostgreSQL (prod) |
| Google AI | Gemini API |
| Deployment | Railway / Google Cloud Run (coming soon) |
- Python 3.11+
- Git
git clone https://github.com/soaebhasan12/OrganBridge.git
cd OrganBridge/organBridgepython -m venv venv
# Windows
venv\Scripts\activate
# macOS / Linux
source venv/bin/activatepip install -r requirements.txtSECRET_KEY=your-secret-key-here
DEBUG=True
GEMINI_API_KEY=your-gemini-api-keypython manage.py migratepython manage.py createsuperuserpython manage.py runserverThe application will be available at http://127.0.0.1:8000/
| Metric | Value |
|---|---|
| Dataset Size | 900+ donor-recipient records |
| Model Accuracy | 95% on test data |
| Compatibility Factors | 15+ per match |
| Matching Speed | Sub-second inference |
- All match details visible only to directly involved donor and recipient
- Personal contact information hidden until match acceptance
- Every sensitive view protected with
@login_required - CSRF protection on all forms
- Environment variables for all sensitive credentials
This project is submitted for the Google Solution Challenge 2026 India — Build with AI
- Theme: Unbiased AI Decision-Making
- Team: Orion-XAI
- Track: Theme 4 — Ensuring Fairness and Detecting Bias in Automated Decisions
This project was built for a hackathon. MIT License — see LICENSE for details.
Pull requests are welcome. For significant changes, please open an issue first.
Built with ❤️ by Team Orion-XAI for Google Solution Challenge 2026