in India, thousand of thalassemia patients- especially children depends on timely bond transfusion to survive many come from underserved areas with limited access to healthcare infrastructure. We were inspired to solve this life critical problem using AI and automation, aiming to simplify the process for even non-tech savvy users and reduce preventable emergencies due to delayed transfusion
The system predicts transfusion urgency using clinical data like Hg, iron and transfusion dates it uses smart OCR extract values from prescription even handwritten ones and update record in real time it matches donors via API (e-RaktKosh,blood banks) and alerts via SMS/WhatsApp it also schedules transfusion and provides automated reminders,enabling hospitals and patients to coordinates seamlessly
we built the backend using Flask-Dance For Google OAuth for ML we trained an ensemble model (XGBoost,CatBoost,Logisits Regression) to predict urgency OCR is powered by tesseract, optimized for medical text data is stored using pandas in CSV formats synced in real time Notification are sent via SMS , via Twilio and SMTP the fronted uses HTML,CSS, and JavaScript with QR- based patient lookup
Extracts accurate values from handwritten prescription Aligning donor availability data from external APIs Ensuring data privacy for patient health information keeping urgency prediction accurate with incomplete patient history
smart OCR Real Time urgency prediction with auto-scheduling and alerts seamless donor patient matching using live data sources
ml model integration we learned through
Integrating hospital EMR system to fetch data directly multi language support for broader accessiblity Turning the project into an open-sources public health tool
(we include "QUANTITATIVE" and "QUALITATIVE" approach)
Log in or sign up for Devpost to join the conversation.