BloodBridge AI – Our Project Story

Inspiration

It all started when we discovered Blood Warriors, an NGO supporting over 3,800 Thalassemia donors through chaotic WhatsApp groups. Every month, they coordinate 6,000+ blood requests, but critical messages often get lost in group chat noise. Lives were literally at stake because of manual coordination failures. They had a concept called the “Blood Bridge” system, where each patient is supported by a chain of 8–10 committed donors. Unfortunately, the system kept breaking due to a lack of automation and smart matching. That’s when it hit us: Why build yet another app when WhatsApp is already the heartbeat of this community?

What We Learned

  1. Work With Existing Habits WhatsApp isn’t just convenient—it's critical infrastructure for this community. Reinventing the wheel wasn't the answer. Enhancing their existing tools was.
    1. Gamification Drives Real Behavior Donors loved seeing how many lives they’ve saved, climbing leaderboards, and unlocking badges. This simple approach increased donor retention by ~60%.
    2. AI + Human Coordination = Best of Both Worlds Automation shines when routing messages and predicting availability. But when it comes to empathy and trust? That’s still human territory. We focused on amplifying, not replacing, the human element.

How We Built It

🔧 Architecture We created a lightweight backend connected to the WhatsApp Business API using a webhook system. It interprets commands, routes emergencies, and tracks donor stats in real-time.

The AI Under the Hood

  1. Intent Recognition (NLP) Powered by fine-tuned multilingual BERT, our model distinguishes between casual messages and structured commands like: /emergency B+ Mumbai
  2. Donor Availability Prediction We use a Random Forest model that scores donors based on: Location match Last donation date Avg. response time Day-of-week availability
  3. Gamification Engine We use a custom scoring algorithm to reward donors based on donation impact, urgency, and timing: Managing the Blood Bridge Chain We ensure every patient has a functioning 8–10 person donor chain. If someone in the chain becomes unavailable, the system finds and alerts a backup.

Challenges We Faced

  1. WhatsApp API Limits Issue: Approval takes time, and there are rate limits. Solution: We prototyped using Twilio Sandbox and added an SMS fallback for emergencies.
  2. Medical Privacy & Security Issue: WhatsApp isn’t built for handling sensitive health data. Solution: No medical records are stored or transmitted. Only coordination data is shared. All messages are end-to-end encrypted, and the architecture aligns with HIPAA best practices.
  3. Understanding Casual Multilingual Messages Issue: Donors often use a mix of English, Hindi, and abbreviations. Solution: We trained our models on actual donor messages and fine-tuned BERT with Blood Warriors’ real-world data.
  4. Speed in Life-Or-Death Situations Issue: Our goal was sub-15 second response time. Solution: We optimized DB queries, used Redis caching, and implemented async parallel messaging.

Hackathon Demo Highlights

🚀 Hackathon Deliverables (24 Hours) Hours 1-6: WhatsApp AI Foundation WhatsApp webhook setup and secure message processing pipeline Core NLP intent recognition (emergency, information, statistics, gamification) Basic donor database schema with Blood Bridge relationship mapping User authentication and registration flow through WhatsApp Hours 7-14: Intelligence & Coordination Engine Blood Bridge coordination logic (manage 8-10 donor chains per patient) Emergency blood request system with automated donor alerts FAQ automation system (top 20 Thalassemia questions with AI responses) Donor availability prediction model using historical patterns Hours 15-22: Gamification Revolution /mystats command: Personal donation statistics and badge tracking /leaderboard functionality: Monthly donor rankings and competitions /badges system: Achievement unlocking and milestone celebrations /challenge feature: Weekly donor engagement missions Hours 23-24: Live Demo Perfection Interactive demonstration scenarios with realistic donor interactions Performance optimization and comprehensive testing Final presentation preparation and live demo setup Deliverable: Fully functional WhatsApp AI chatbot with gamification - Interactive live demo ready!

#What’s Next Post-Hackathon Strategic Milestones Month 1-2: Pilot Launch & Validation Month 3-4: Scale & Integrate Ecosystem Month 5-8: Advanced Intelligence & Features Month 9-12: National Expansion & Impact

Built With

Share this project:

Updates