HealthIA

Azure Cloud Architecture
Agentic Framework
Diet Generator
Meal Scanner with Open AI
Workout Planner
Medical Personalization

Inspiration

Watching friends and family struggle with diabetes, obesity, and stress showed us that generic health apps rarely stick. We envisioned an always-on coach that adapts to real-time biosignals and makes healthy living feel effortless.

What it does

HealthIA ingests data from wearables—smartwatches and smartbands tracking heart rate, sleep, and steps—and combines it with chat input and dish photos to deliver hyper-personalised meal plans, workouts, and habit nudges. Light gamification (streaks + badges) keeps users motivated, turning raw metrics into sustainable daily routines.

How we built it

Java + Spring Boot API exposes /api/v1/chatbot and other endpoints.
LangGraph multi-agent flow (Planner → Retriever → Generator) orchestrates expertise from Medical, Nutrition, and Fitness agents.
OpenAI GPT-4o (text + image) powers conversation, reasoning, and dish recognition.
Kafka/WebSockets ingestion layer streams wearable data with millisecond latency.
Azure Database for MySQL stores user profiles, metrics, and vector embeddings.
Progressive Web App pushes recipes, workouts, and real-time alerts.
CI/CD via GitHub Actions ships containers to cloud runtime in minutes.

Challenges we ran into

Encrypting sensitive health data without adding noticeable lag.
Maintaining safe, medically sound advice while keeping the chat friendly.
Achieving reliable real-time updates on patchy mobile networks.
Packing full IoT, vision, and chat integration into a six-week hackathon window.

Accomplishments that we're proud of

Live demos showing meal and workout adjustments within seconds of a heart-rate spike.
A coaching experience that feels genuinely personalised.
A modular multi-agent architecture that new specialists can plug into easily.
Early testers hitting 20-day streaks—double the typical health-app retention.