JeevanSetu-Smart donor matching for stemcell transplantation

JeevanSetu : Proactive Thalassemia care. We use Time-Series AI to predict blood needs & OCR/NLP for HLA matching. GenAI-driven retention transforms reactive crisis into predictive genetic precision.

Comment

Inspiration

Thalassemia patients often live in a "panic-mode" cycle, where care is reactive rather than planned. We saw families struggling with last-minute blood searches and the agonizingly slow process of manual HLA matching for curative transplants. We built JeevanSetu (Bridge to Life) to transform this fragmented struggle into a proactive, AI-driven ecosystem that anticipates needs before they become crises.

What it does

JeevanSetu is a proactive care platform that:

  • Predicts Transfusion Needs: Uses Time-Series AI to forecast hemoglobin drops, allowing patients to pre-schedule transfusions.
  • Automates Genetic Matching: Employs OCR & NLP to instantly parse 10-point HLA alleles from unstructured lab reports, replacing weeks of manual coordination.
  • Boosts Donor Retention: Leverages GenAI to create personalized "Impact Narratives," building emotional connections between donors and patients.
  • Connects the Ecosystem: Synchronizes patients, donors, and clinicians via a real-time, geo-spatial dashboard.

How we built it

We engineered a robust, privacy-first stack:

  • Intelligence: Python-based core using TensorFlow/Scikit-learn for predictions and HuggingFace for clinical NLP.
  • Backend: FastAPI and Node.js orchestrating data across PostgreSQL (clinical records) and MongoDB (unstructured logs).
  • Frontend: A cross-platform Flutter mobile app and React.js admin dashboard integrated with Google Maps API.
  • Security: Built with HIPAA-compliant encryption and ABDM-standard interoperability gateways.

Challenges we ran into

One major hurdle was the "Cold-Start" problem—predictive models need historical data to be accurate. We addressed this by implementing transfer learning from generalized datasets. We also navigated the Legacy Interoperability gap, where rural centers lack standardized APIs, requiring us to design an architecture that can eventually bridge these digital "blind spots."

Accomplishments that we're proud of

We are incredibly proud of our HLA Parser. Transforming a messy, high-stakes medical PDF into a structured genetic profile in seconds—a process that usually takes clinicians days of manual entry—was a massive technical win. Creating a unified "Google Maps for Life" that feels seamless for the end-user was equally rewarding.

What we learned

We gained deep expertise in HL7 FHIR & ABDM protocols for healthcare interoperability. More importantly, we learned that AI in medicine shouldn't replace doctors; it should act as a high-precision Decision Support System that handles the data heavy-lifting so clinicians can focus on saving lives.

What's next for JeevanSetu

Our next milestone is tackling Last-Mile Connectivity for rural centers using "Lite" SMS-based notifications and offline-first data syncing. We also plan to integrate with national blood registries to expand our donor pool, making the "Bridge to Life" accessible to every Thalassemia patient, regardless of their location.

Built With

Share this project:

Updates