Inspiration
Managing diabetes feels like a 24/7 grind, constantly logging food, guessing carb counts, and stressing about what one meal might do to your body long-term. We wanted to make this way easier and actually cool by using AI to instantly show how food affects not just your glucose, but also your brain and overall health.
What it does
PlateSense is an AI-powered tool where you just snap a photo of your food and boom, you get instant predictions on glucose spikes and health risk levels. It syncs your phone to your computer in real time using a QR code (no app download), and builds personalized 7-day diet and workout plans. The coolest part? A 3D brain visualizer that shows how nutrients impact real brain areas like the Prefrontal Cortex and Hippocampus, so you can actually see how food affects your mind.
How we built it
Frontend: React 18 + Tailwind CSS, with Three.js (React Three Fiber) to power the interactive 3D brain.
Backend: Node.js and Express, with MongoDB storing user data and meal history.
Real-time Sync: Socket.IO streams photos straight from your phone camera to the desktop dashboard instantly.
AI: Gemini 2.0 Flash Lite (via OpenRouter) handles food image analysis and personalized health plans.
Challenges we ran into
Cross-device syncing: Making phone-to-desktop camera sharing feel instant using WebSockets and session IDs.
3D brain mapping: Turning AI nutrition data into accurate hotspots on a 3D brain model.
Data accuracy: Getting consistent glucose predictions across different users with things like BMI and insulin ratios.
Accomplishments we’re proud of
Building a zero-friction mobile experience with no app store download required.
Creating an AI system that turns one food photo into instant health insights and long-term plans.
Designing a visual brain-health experience that makes something super abstract feel real and understandable.
What we learned
How to manage real-time state across multiple devices using WebSockets.
Why structured AI outputs are super important for charts and 3D visuals.
How to securely handle authentication and file uploads in a health-based app.
What’s next for PlateSense
Wearables: Connecting Continuous Glucose Monitors (CGMs) to compare predictions with real data.
Community: Letting users share safe meals and improve recipes together.
Brain upgrades: More detailed cognitive metrics and long-term brain health tracking over time
Mobile app: shows the user their data without the use of a laptop/computer.
Log in or sign up for Devpost to join the conversation.