🩺 MediSync

"Every signal. One story. Your health, finally connected."

💡 Inspiration

Today’s healthcare data is deeply fragmented. Your doctor sees your vitals once a year, your fitness tracker monitors your daily steps, your pharmacy logs your medications, and your insurer tracks your claims. Because none of these systems communicate with each other, the person with the least complete picture of their own health is the patient. We built MediSync to fix this by giving everyday people a unified, explainable command center for their health.

🚀 What it does

MediSync is a personal health intelligence platform that pulls disparate health signals into a single unified layer. Rather than just displaying data, it tells you what that data means together.

Smart Intelligence Engine: A correlation engine that cross-references data streams to generate explainable insights. For example, it can link your wearable sleep data to your step count, or correlate your logged diet with reported symptoms.
Comprehensive Tracking: It aggregates wearable data, medication schedules, doctor visits, family health history, and nutrition logs.
Actionable Nudges: Every insight provides the "What", the "Why" (source signals), a "Confidence" score, and a clear "Action" the user can take next.

🛠️ How we built it

We developed MediSync during a 48-hour sprint with a four-person team divided into specialized frontend, backend, intelligence, and full-stack feature roles.

Tech Stack

Frontend: React (PWA), Tailwind CSS, Zustand, Recharts
Backend: Node.js, Express
Database: SQLite (via Knex)
AI Layer: Claude API (via Anthropic)

API Integrations

openFDA: Drug interactions and recall alerts
Infermedica: Symptom checking and condition autocomplete
NPPES & CMS.gov: Provider search and quality ratings
Nutritionix & FoodData Central: Food search and nutrient profiles
disease.sh: Outbreak awareness and tracking
Healthcare.gov: Insurance marketplace plan browsing

⚠️ Challenges we ran into

Integrating over six external APIs simultaneously was our biggest hurdle. To prevent merge conflicts across our four-person team, we established strict JSON response contracts and shared TypeScript definitions in the first 30 minutes of the hackathon.

Additionally, managing external API limits required defensive engineering. We implemented an in-memory caching layer with TTLs, a rate limit handler (especially for openFDA's 240 calls/minute limit), and an API fallback layer to gracefully handle third-party downtime.

🏆 Accomplishments that we're proud of

We successfully built a functional cross-signal correlation engine capable of evaluating complex, multi-variable health rules in real-time. We're especially proud of the engine's ability to generate "Wow moment" insights—such as recognizing a family history of diabetes, pairing it with a user's rising resting heart rate, and subsequently recommending a nearby endocrinologist.

We also strictly adhered to a "Data Minimalism" ethical framework: collecting only what is necessary, avoiding irrelevant data (like social media history), and ensuring total transparency by showing the user exactly which data points generated a specific insight.

📚 What we learned

We learned that the true value of health tech isn't in collecting more data, but in building relationships between the data points we already have. Technically, we learned the immense value of defining strict API contracts before writing any logic, which allowed our frontend and backend tracks to work completely in parallel.

🔮 What's next for MediSync

Our immediate next step is swapping our mock wearable WebSocket simulator for real-world integrations. We also plan to expand our notification queue to support external push notifications for time-sensitive nudges and further refine our composite Health Risk Score algorithm to drive better daily user engagement.