Sanjeevani Setu by Garuda.AI

They are truly wise who strive to save lives

Comment

🚀 Inspiration

In 2021, we met Ananya, a 10-year-old girl from a village near Varanasi who suffers from Thalassemia Major, a chronic blood disorder. She requires a transfusion every 15–20 days. But due to poor infrastructure and a lack of available blood, her parents often travel over 60 km only to be turned away. Her treatment is sometimes delayed by days — resulting in fatigue, infections, and missed school. She’s not alone.

🧬 The Reality

  • India requires 12 million units of blood each year, but only collects around 9 million, leading to a shortfall of 3 million units annually.
    Source: Ministry of Health & Family Welfare (MoHFW), 2023

  • Despite government efforts, only 1% of India’s population donates blood, far short of the 3–4% needed to meet national demand.
    _Source: WHO, 2022

  • In rural areas, awareness and participation are critically low. A 2020 study by AIIMS found that 62% of rural citizens had never donated blood, mainly due to lack of awareness or access.

  • Over 100,000 children in India live with Thalassemia, requiring regular transfusions — sometimes every 10 days. Missed transfusions can lead to organ failure or death.
    Source: Indian Academy of Pediatrics (IAP), 2022

  • 40% of blood donations in India happen in emergencies, not proactively — which means patients with chronic conditions are at higher risk of delays or unavailability.
    Source: NACO Annual Report, 2021

These gaps are not just statistical — they are life-threatening.

❗ Why Awareness Alone Isn’t Working

While numerous blood donation camps and government campaigns have improved overall donations, there is no intelligent system that:

  • Detects who needs blood urgently based on disease severity
  • Predicts when a donor is most likely to donate again
  • Triggers real-time blood alerts to verified, eligible donors
  • Connects blood with demand before a crisis occurs

Ananya’s story, along with these systemic failures, moved us to build Sanjeevani Setu — a platform that predicts, prioritizes, and mobilizes blood based on real medical urgency, not just availability.

Blood is not just about donation; it’s about delivering life where and when it’s needed most. And we believe AI can make that happen.

💡 What it does

Sanjeevani Setu is a real-time, AI-powered blood allocation and emergency response system built for patients with blood disorders like Thalassemia, Sickle Cell Anemia, Leukemia, and trauma cases.

Key features:

  • 🧠 AI-based health record analysis to detect high-risk patients automatically
  • 📡 Smart geo-matching system to locate blood within a 25–50 km radius
  • 🔁 Emergency outreach module to alert eligible donors via calls, WhatsApp, SMS, and email
  • 🔒 Full integration with e-RaktKosh, Blood Warriors’ systems, and hospital records using secure APIs
  • 📲 Patient & doctor dashboard to track transfusion status, requests, and delivery ETAs

The system guarantees blood delivery within 1 to 1.5 hours for critical patients.

🛠️ How we built it

We used a combination of open-source tools and Microsoft Azure services.

Tech Stack:

  • Frontend: React.js + TailwindCSS for mobile/web dashboard
  • Backend: FastAPI (Python) for API orchestration and logic
  • ML Models:
    • XGBoost for patient severity classification
    • ARIMA for blood requirement forecasting
  • NLP: Azure Language Studio + custom LLM for scanning unstructured EHR notes
  • Mapping & ETA: Azure Maps + Google Distance Matrix API
  • Data: Azure SQL Database + MongoDB
  • Integrations: FHIR APIs (Ayushman Bharat), e-RaktKosh, Blood Bridge (mock APIs)
  • Messaging: Twilio for SMS/Calls, WhatsApp Business API for alerts

We applied logistic regression on sample patient records to detect anemia risk and used clustering (KMeans) to optimize donor location zones.

Challenges we ran into

  • Privacy & Access: Getting access to health records while respecting data protection laws was a major challenge. We simulated EHRs using dummy data conforming to FHIR standards.
  • Reliable real-time geolocation: Ensuring timely donor-patient matching with low latency, especially in remote areas with limited connectivity.
  • API compatibility: Integrating with existing platforms like e-RaktKosh required understanding their architecture and simulating responses.
  • Donor responsiveness: Building a behavioral model to predict whether a donor would respond to an alert was complex due to lack of historical data.

🏅 Accomplishments that we're proud of

  • Clearly defined the end-to-end system architecture for an AI-powered blood allocation and disease-prioritized response platform.
  • Designed a detailed feature set and user workflow, addressing real-world challenges faced by Thalassemia and critical care patients.
  • Conducted extensive research on healthcare gaps, blood donation behavior, and existing systems like e-RaktKosh to validate our idea.
  • Created a data model and decision flow for prioritizing patients based on severity, location, and urgency.
  • Drafted a communication strategy for contacting donors via WhatsApp, SMS, and calls using AI-driven triggers.
  • Outlined the full technology stack, including integration points for hospitals, blood banks, and NGOs.

With limited time, we focused on building a rock-solid foundation for a scalable solution that can be implemented in future hackathon phases or pilot deployments.

📚 What we learned

  • How to structure and process medical records using FHIR and HL7 standards
  • Application of machine learning for real-time risk stratification
  • Usage of location intelligence and delivery optimization models for health infrastructure
  • Importance of ethical AI when working with life-critical data

We also learned that technology must be human-centered — a simple voice message to a donor can mean the difference between life and death.

🛤️ What's next for Sanjeevani Setu by Garuda.AI

  • 🧪 Pilot with NGOs & hospitals: Test the system with real patients under guidance of Blood Warriors and similar organizations.
  • 🧬 Expand to cancer care, surgery, and pregnancy-related blood emergencies
  • 📡 Integrate IoT-based cold chain tracking for sensitive blood product delivery
  • 🏥 Partner with state governments to enable 24/7 blood response units
  • 🤖 Build a self-learning model that improves donor outreach timing and language using reinforcement learning

We believe Sanjeevani Setu can become India’s digital backbone for equitable, intelligent, and timely blood distribution.

“Blood is not just a donation — it’s data, it’s logistics, it’s life. We aim to transform it all with AI.”

Built With

  • and
  • arima
  • azure
  • azure-language-studio
  • azure-machine-learning-studio
  • azure-maps
  • bert-(for-nlp)
  • blob-storage
  • e-raktkosh-api
  • fastapi
  • fhir-api-(hl7)
  • github
  • google-distance-matrix-api
  • lstm-(for-time-series)
  • maps
  • microsoft-azure-(functions
  • ml-studio
  • mongodb
  • postman
  • python
  • react.js
  • scikit-learn
  • spacy
  • sql-database)
  • tailwind-css
  • twilio
  • vs
  • whatsapp-business-api
  • xgboost
Share this project:

Updates