Inspiration
Every year, thousands of lives are lost due to delays in finding compatible blood donors, especially in rural and under-resourced regions. The fragmented and outdated infrastructure in blood donation services inspired us to create a smart, secure, and scalable system that ensures real-time blood matching, while protecting donor data and streamlining medical workflows. We were motivated by both personal experiences and the critical need for decentralized, AI-driven solutions that can reach people even in low-connectivity areas.
What it does
Hemoglobe AI is an end-to-end intelligent blood management ecosystem that:
Matches patients with donors in real-time using AI.
Provides predictive analytics on future blood demands.
Uses blockchain to ensure secure and tamper-proof donor data.
Offers telemedicine consultation for remote areas.
Integrates with e-RaktKosh, Blood Bridge, and UHI networks for government-level interoperability.
Works both online and offline using a Progressive Web App (PWA) model.
How we built it
We used a layered architecture:
Presentation Layer Mobile App: React Native + Redux
Web Portal: Vue.js + Vuetify
Admin Dashboard: Angular Material
Service Layer API Gateway: Node.js + Express
AI Matching Engine: Python (TensorFlow)
Predictive Analytics: PyTorch, Prophet
Telemedicine: WebRTC + Twilio
Data Layer Database: Firebase Firestore
Blockchain: Hyperledger Fabric for donor data
Medical Records: HL7 FHIR protocol
External Integrations e-RaktKosh API, Blood Bridge, UHI
Challenges we ran into
Ensuring data privacy with distributed learning methods (Federated Learning).
Integrating legacy healthcare APIs with modern web architecture.
Creating a responsive system that works well in low-network regions.
Balancing blockchain decentralization with real-time performance.
Training AI models efficiently using Edge Computing for regional deployments.
Accomplishments that we're proud of
Successfully built a working prototype that supports real-time donor-patient matching.
Integrated a hybrid blockchain network for transparency and security.
Enabled teleconsultation within the same ecosystem.
Implemented predictive models that forecast demand for blood by location.
Designed a microservice-based architecture ready for scaling across regions.
What we learned
Deep understanding of healthcare protocols like HL7 FHIR.
Real-world use of Federated Learning in privacy-sensitive domains.
Practical integration of AI, blockchain, and telemedicine into a single unified system.
How to build and deploy microservices with seamless cross-platform communication.
Importance of offline-first architecture in global health contexts.
What's next for Hemoglobe AI System Architecture
Integration with national health registries and insurance systems.
Scaling the platform across multiple states or countries with different health infrastructures.
Implementing voice-based support for non-tech-savvy users.
Enhancing AI engine to include platelet, plasma, and rare blood type forecasting.
Publishing our work as a whitepaper and contributing to open-source healthcare AI communities.
Built With
- amazon-web-services
- docker
- express.js
- firebase
- google-cloud
- hyperledger
- javascript
- jwt
- node.js
- python
- pytorch
- react-native
- redux
- tensorflow
- typescript
- vue.js
- vuetify
Log in or sign up for Devpost to join the conversation.