Swasthya-Setu is an AI-powered health triage and population intelligence platform designed to solve last-mile healthcare challenges.
It enables medical interns and field health workers to digitize patient data, upload medical reports, and receive AI-assisted clinical insights while simultaneously contributing to large-scale public health intelligence.
The platform acts as a “Field Hospital in a Pocket” by combining individual-level care with community-level outbreak monitoring.
Healthcare delivery in rural and semi-urban regions suffers from:
- Paper-based and fragmented records
- No longitudinal patient history
- Delayed disease detection
- Manual and delayed outbreak reporting
As a result:
- Preventable diseases escalate
- Hospitals get overloaded
- Government health spending becomes reactive instead of proactive
Swasthya-Setu provides an end-to-end AI-driven healthcare intelligence pipeline:
- Digitization of medical reports, vitals, and symptoms
- AI-based extraction using vision and language models
- Creation of longitudinal patient health profiles
- Real-time medical triage and risk scoring
- Anonymized aggregation for outbreak and trend detection
- Unified digital health records
- Auto-extraction from prescriptions, lab reports, and scans
- Context-aware AI responses using past medical history
- Symptom and report-based risk assessment
- Urgency classification (Low / Medium / Critical)
- Clinical decision support for interns and doctors
- Medical LLMs for reasoning and summarization
- Vision models for document and scan understanding
- Retrieval-Augmented Generation (RAG)
- Ensemble inference for higher robustness
Swasthya-Setu continuously analyzes anonymized, area-wise health data collected by field workers.
Using temporal and geo-spatial pattern analysis, the system detects abnormal spikes in symptoms and vitals, acting as an early warning system for potential disease outbreaks.
- Faster outbreak identification
- Proactive public health response
- Smarter allocation of government resources
- Reduced hospital overload
- Area-wise disease trend analysis
- Integration with Anganwadi and grassroots data
- Evidence-based decision support for policymakers
- Data-driven optimization of healthcare spending
- React / React Native (Expo)
- Tailwind CSS / ShadCN UI
- Offline-first design for field environments
- Node.js and Express
- FastAPI for AI inference services
- PostgreSQL and Vector Databases (FAISS / Qdrant)
- Medical-tuned LLMs (LLaMA / Mistral via Ollama)
- Vision Transformers for medical document analysis
- Time-series and geo-spatial analytics
- Secure health data storage
- JWT-based authentication
- Role-based access control
- Privacy-preserving data aggregation
Intern / Field Worker
→ Patient Registration
→ Vitals & Report Upload
→ AI Data Extraction
→ Patient Health Profile
→ AI Triage & Insights
→ Regional Analytics & Alerts
- Early disease detection
- Empowerment of medical interns
- Improved healthcare access in underserved regions
- Smarter public health planning
- Scalable to national healthcare programs
- Core architecture designed
- AI pipelines prototyped
- MVP under active development
- Hackathon-ready with scalable vision
To become a national-scale digital health intelligence layer that bridges grassroots healthcare data with AI-powered decision-making—ensuring healthcare reaches people before emergencies arise.
This project is developed as part of a initiative focused on AI for social good and healthcare innovation.
Ideas, feedback, and collaboration are welcome.