ThalaConnect AI – Hackathon Submission
Inspiration
Thalassemia is more than just a medical condition—it’s a lifelong struggle of appointments, uncertainty, and isolation for millions around the world. Our team was inspired by the lived experiences of patients and caregivers who battle not just the disease, but also the fragmented and inefficient support systems around them. We wanted to build a solution that would empower this community through intelligent, empathetic, and secure digital support—something that could truly make a difference in their daily lives.
What it does
ThalaConnect AI is an all-in-one, AI-powered digital ecosystem tailored specifically for the Thalassemia community. It consists of three core modules:
ThalaGuide (AI Knowledge & Lifestyle Hub): A conversational RAG chatbot and Food Architecture Agent that answers complex medical queries and generates custom, Thalassemia-safe recipes using LLMs.
CareLink (Secure Communication Portal): A HIPAA/DPDP-compliant encrypted messaging and appointment booking system for seamless communication with healthcare providers.
Warrior's Circle (Anonymous Community Forum): A safe, moderated, pseudonymous platform where patients and caregivers can share experiences and find emotional support.
Together, these components form a compassionate, practical, and secure assistant for Thalassemia patients, caregivers, and providers.
How we built it
We adopted a Microservices Architecture to keep each component independent, scalable, and maintainable.
- Backend (Go): Developed the core application logic for user authentication, messaging, and community forums.
- AI/ML Services (Python): Created microservices using Flask to handle embeddings (with sentence-transformers), recipe scraping (with BeautifulSoup), and LLM interactions (via Gemini 1.5 Flash).
- Database: Used PostgreSQL with the
pgvectorextension for structured data and vector search. - Frontend: Built with React for a responsive and intuitive user interface.
- Deployment: Containerized using Docker and deployed on Microsoft Azure using AKS and Blob Storage.
- Security: Implemented E2EE messaging and moderation with a combination of Go services and AI scanning tools.
Challenges we ran into
- Time Constraint: With only 48 hours, prioritizing features and scoping the MVP was critical.
- LLM Safety & Hallucination: We needed to ground LLM responses using a RAG pipeline to ensure accuracy, particularly for medical information.
- Web Scraper Reliability: Adapting scraping logic quickly for recipe sites while avoiding structural dependencies was tricky.
- Data Privacy: Ensuring secure and private handling of sensitive health data required rigorous attention to architecture and encryption methods.
Accomplishments that we're proud of
- Developed a fully working AI-powered recipe personalization pipeline grounded in nutrition and medical safety.
- Integrated LLM-based RAG chatbot grounded on real, vetted Thalassemia documents.
- Implemented end-to-end encrypted chat with appointment scheduling features.
- Created a live, functioning pseudonymous community forum with AI moderation.
- Designed a platform that feels empathetic, not just functional—something we believe patients will actually want to use.
What we learned
- How to efficiently integrate LLMs (Gemini) into a RAG pipeline with real-world constraints.
- The power of microservices in rapidly iterating and scaling different platform features in parallel.
- Real-world application of secure messaging and privacy-focused design in healthcare.
- The importance of user-centered design, especially when dealing with chronic illness communities.
- That small innovations—like personalized recipes—can have a massive emotional and psychological impact.
What's next for ThalaConnect AI
- Mobile App: Build a fully native mobile experience for patients on the go.
- Multi-language Support: Add regional language capabilities to serve non-English speaking users across India and beyond.
- Real-Time Health Integrations: Integrate with wearable devices and lab systems for better health tracking.
- Nutrition Database Expansion: Partner with nutritionists and integrate a broader API to generate even more localized and diverse recipes.
- Mental Health Support Module: Add AI-guided journaling and access to counseling resources.
- Official Deployment: Collaborate with the Blood Warriors Foundation to onboard real patients and launch pilot testing.
Tagline: ThalaConnect AI — Empowering Every Thalassemia Journey with Knowledge, Care, and Community.
Built With
- 1.5
- access
- aks)
- api
- apps
- azure
- beautiful-soup
- blob
- compose
- container
- control
- docker
- e2ee)
- edamam
- encryption
- end-to-end
- flash
- flask
- gemini
- git
- github
- go
- golang)
- javascript
- jwt
- kubernetes
- libraries
- microsoft
- nginx
- nutrition
- pgvector
- postgresql
- postman
- python
- react
- react)
- requests
- role-based
- sentence-transformers
- service
- storage
Log in or sign up for Devpost to join the conversation.