🩸 About the Project – LifeStream

🔍 Inspiration

Every year, India faces critical blood shortages 🚨. Despite rising smartphone and UPI usage, voluntary blood donation remains low due to app fatigue, donor privacy concerns, and fragmented hospital workflows 🏥.

We realized: people want to help—but the process must be invisible, embedded, and effortless.

That’s how LifeStream was born—an AI-powered blood donation system built into UPI, WhatsApp, and banking apps that people already use daily 💡. No installs, no friction—just life-saving action, triggered by everyday habits.

🛠️ How We Built It

Frontend Simulation 🧾: Designed a mock UPI payment screen that displays a clean opt-in prompt during payment, asking users if they’d like to join the donor network.
Donor Registration Logic 🩸: Simulated Aadhaar-linked health data autofill (e.g., blood type), and stored user consents and preferences securely.
Hospital Request Interface 🏥: Built a basic voice/text input tool for hospitals using Whisper and GPT-4o to turn natural language into structured blood requests.
AI Matching Engine 🤖: Created a context-aware donor matching system using donation frequency, proximity, availability, and urgency scoring (XGBoost + Graph Neural Networks).
WhatsApp Bot 📲: Integrated WhatsApp Business API sandbox to send live alerts to matched donors and capture quick replies (Y/N).
Privacy-first Architecture 🔒: Simulated on-device processing and used placeholder homomorphic encryption to ensure privacy compliance.
Rural IVR Support 📞: Designed SMS/IVR fallback logic for users without smartphones or internet access.

📚 What We Learned

Friction = Failure 🚫: Users resist separate apps for infrequent tasks. Embedding civic tasks into daily flows increases participation.
Conversational Interfaces Win 💬: WhatsApp + voice made onboarding smooth even for low-tech users.
Context Matters ⏰: Timing, location patterns, and donor history improved matching accuracy and urgency response.
Privacy by Design is Non-Negotiable 🔐: Data trust drives adoption. We prioritized anonymization and user-controlled sharing.

⚠️ Challenges Faced

UPI & Aadhaar Access 🏦: Due to restricted API access, we simulated these parts but designed them modularly for easy future integration.
WhatsApp Sandbox Limitations 🧪: Couldn't fully test live flows, so logic was built using dummy datasets and scripted responses.
Balancing Personalization & Privacy ⚖️: Designing context-aware alerts while preserving anonymity was complex.
Rural Accessibility 🏡: Ensuring inclusivity via SMS/IVR meant planning for low-bandwidth and low-digital-literacy environments.