Inspiration:
Thalassemia patients require frequent blood transfusions, often every 2–4 weeks. Yet, despite digital initiatives like e-RaktKosh, patients still face delays and uncertainty, especially in rural or low-networked areas. Our inspiration came from real stories where families had to post blood requests on social media during emergencies.
We desired to build a platform that predicts donor availability, builds supportive communication between doctors and patients, and encourages recurring donations—all in real-time and at scale. This gave birth to ThalConnect.
What it does:
ThalConnect is a conceptual AI-powered ecosystem that:
1)Matches patients to suitable blood donors in real-time using location, blood group, and availability 2)Predicts future donor availability based on time-series analysis 3)Encourages recurring donations through a gamified dashboard 4)Facilitates communication between patients, caregivers, and hematologists 5)Provides multilingual education materials for Thalassemia care 6)Maintains data privacy using secure, encrypted, consent-driven access
How we built it:
We structured the system with 6 modular components: 1)Donor Matching Engine using geolocation, last donation date, blood group. 2)AI Prediction Model: 𝑃(𝐷(𝑡+Δ𝑡)=1∣history, preferences, time) built using historical donation logs and behavior patterns. 3)Gamified Donor Portal to track milestones, show impact, and schedule next donations. 4)Patient-Doctor Communication Module for scheduling, chat, and alerts. 5)Education Center with interactive chatbot and regionally translated health guides. 6)Security Layer with AES encryption, OAuth2, RBAC, and FHIR compliance.
Challenges we ran into:
1)Donor Prediction with Sparse Data: Donor behavior isn’t always consistent. We proposed modeling using seasonal patterns and voluntary inputs. 2)Legacy Integration: Systems like e-RaktKosh weren’t built for AI integration. We planned to use middleware and API caching for stability. 3)Motivation and Engagement: Many one-time donors need encouragement. We used gamification and impact visualizations to build a habit. 4)Language Barriers: Patients may not be literate or tech-savvy. We designed multilingual visuals and chatbot flows to support education.
Accomplishments that we're proud of:
1)Designed an end-to-end AI-first architecture that supports both patient-side urgency and donor-side empathy 2)Created a realistic model for donor availability prediction 3)Incorporated gamification and emotional storytelling into donor tracking 4)Ensured healthcare-grade data privacy from Day 1 5)Designed with interoperability to plug into existing systems like e-RaktKosh
What we learned:
1)AI isn’t always about models—user behavior and incentives matter just as much 2)The need for human-centered design in healthcare systems is critical 3)Government-backed systems are powerful, but integration needs patience and planning 4)Even in data-limited settings, probabilistic models and pattern recognition can make a real difference
What's next for ThalConnect
1)Develop a working prototype integrating mock APIs from e-RaktKosh and donor databases 2)Deploy AI model using Prophet/XGBoost to simulate donation likelihoods 3)Build a mobile-first UI for both patients and donors 4)Partner with NGOs like Blood Warriors to pilot in high-need districts 5)Expand to support other transfusion-dependent diseases
Built With
- amazon-web-services
- e-raktkosh
- flask
- google-maps
- https
- mongodb
- native
- numpy
- oauth2
- pandas
- postgresql
- python
- react
- react.js
- scikit-learn
- tailwind
- xgboost
