Built for HACKHAZARDS '26
Every family deserves a health safety net. SwaasthyaGraph builds it — one graph at a time.
SwaasthyaGraph is a multilingual family health and drug interaction network built to protect families from dangerous medical oversights.
In rural and semi-urban areas, fragmented medical records lead to adverse drug events. A grandfather might be prescribed Warfarin, and later given Aspirin by a different doctor, causing a severe bleeding risk. SwaasthyaGraph solves this by modeling the entire family's health data as a live graph database.
By allowing users to log symptoms and medications via native language voice input (Hindi, Tamil, Telugu), we remove the literacy and language barriers to entry. Background pipelines instantly analyze the data, traverse the family graph to find dangerous drug-drug interactions or hereditary risks, and dispatch SMS alerts.
- 🕸️ Live Family Graph: Visualizes parents, children, and grandparents as nodes, linked to their medications and symptoms.
- 🗣️ Multilingual Voice Logging: Users simply speak in their native tongue (e.g., "Mujhe sir dard hai"). The AI automatically transcribes, translates to English, extracts the medical entities, and injects them into the graph.
⚠️ Drug Interaction Engine: Traverses the graph to find conflicting medications across the family history.- ⚙️ Durable Async Pipelines: Background jobs ensure that complex ancestral risk checks and SMS dispatching never fail.
- 📱 Mobile-First Client: Built natively to feel like an app for end-users, while administrators view the rich dashboard.
This project was specifically architected to leverage the power of four phenomenal partner technologies:
Neo4j acts as the core intelligence engine. Traditional relational databases struggle with complex, multi-generational medical inheritance queries. We use Neo4j AuraDB to natively model CHILD_OF, PRESCRIBED, and EXPERIENCING relationships. Cypher queries can traverse the graph in milliseconds to find active drug contraindications.
To make healthcare accessible in India, English typing cannot be the default. We integrated Sarvam AI to handle the end-to-end voice pipeline: speech-to-text -> translation -> entity extraction (identifying symptoms vs. drugs).
Medical alerts cannot fail. When new symptoms are logged, a webhook triggers a durable 4-step Render Workflow:
- Analyze patient health metrics.
- Scan the Neo4j graph for active drug conflicts.
- Check ancestral hereditary risk paths.
- Compile a clinical summary and dispatch an SMS alert.
The entire client side is built using Expo for cross-platform availability. The dashboard features an embedded Expo web simulator to demonstrate the native mobile feel of the patient-facing application.
Used for rapid API routing configurations to link Sarvam AI output triggers to Neo4j graph mutations.
- Node.js (v18+)
- Neo4j AuraDB instance
- Sarvam AI API Key
- Render account
# Clone the repository
git clone https://github.com/swatiicfai/SwaasthyaGraph.git
cd SwaasthyaGraph
# Install dependencies
npm install
# Start the development server
npm run devThe application will be available at http://localhost:5173.
This project was built with ❤️ over the course of the hackathon by Swati Gupta. We proudly submit this to the HealthTech & Bio Platforms and Human Experience & Productivity themes.