Inspiration
Every year, thousands of children in India are born with Thalassemia, a genetic blood disorder that demands regular blood transfusions for survival. Families struggle to find compatible blood donors in time due to a fragmented and inefficient donation ecosystem. We were inspired to create Blood Warriors AI System to empower organizations and individuals working at the grassroots level—the "Blood Warriors"—with the power of AI to save lives faster and smarter.
What it does
Blood Warriors is an AI-powered blood donation and transfusion management system that:
- Predicts donor availability using ensemble ML models (XGBoost, LSTM, Random Forest)
- Connects patients with suitable donors in real-time
- Forecasts future demand and donor trends
4.Plans for automated notifications, smart reminders, and donor engagement
- Offers a rich dashboard with analytics, metrics, and visual insights 6.Enables recurring donor tracking and engagement loops ## How we built it We designed a scalable full-stack architecture powered by advanced AI: Frontend: React 18 + Tailwind CSS, with rich charts and UI components Backend: Flask (Python 3.11) RESTful API with SQLite (dev) and PostgreSQL (prod) AI/ML Models: XGBoost for fast structured data prediction LSTM for temporal donor behavior Random Forest for feature robustness Ensemble (weighted voting) for accuracy (~73.5%) Setup & Scripts: Cross-platform .bat scripts for installation and execution Deployment Ready: API documented, Docker/Kubernetes compatible
Challenges we ran into
- Lack of real-world data: We had to use high-quality synthetic data due to privacy and availability issues.
- Model generalization: Ensuring models perform well across regions, seasons, and diverse donors.
Healthcare compliance: Designing for HIPAA and Indian data protection laws, even before production. 4.LLM Integration: Early Gemini AI integration was promising but required rollback for stability during hackathon.
Accomplishments that we're proud of
1.Achieved real-time donor matching under 30 seconds
73.5% accuracy from ensemble prediction on synthetic data
Developed a full interactive dashboard with forecasts, model metrics, and donor insights
Designed a system with modular backend, scalable frontend, and integrated AI pipelines 5.Future-ready integration points with e-RaktKosh, LLMs, and wearables
What we learned
AI in healthcare must be interpretable, privacy-compliant, and continuously improving Importance of data engineering and synthetic generation when real-world data is sensitive Challenges of aligning AI predictions with real-time, life-impacting decisions Building for accessibility, reliability, and performance simultaneously
What's next for Blood Warriors AI System
- Develop mobile apps for iOS and Android
- Integrate real-time email/SMS notifications for emergencies
- Deploy Gemini AI for smart donor communication, FAQ generation, and dynamic awareness content
- Add geospatial donor mapping for blood drive planning
- Scale to multi-region deployments with PostgreSQL and cloud services (AWS/GCP) 6.Launch public awareness campaigns and connect with NGOs and healthcare providers
Built With
- docker-ready-other:-gemini-ai-(future)
- gunicorn
- javascript
- jwt-(planned)
- languages:-python-3.11+
- npm
- numpy-deployment-tools:-git
- pandas
- pip
- postgresql-(prod)-machine-learning:-tensorflow-(lstm)
- recharts-backend:-flask
- redis
- rest-apis
- scikit-learn-(random-forest)
- shadcn/ui
- sqlalchemy-databases:-sqlite-(dev)
- tailwind-css
- typescript-frontend:-react-18
- xgboost