Who we are
Movemend is building an AI-powered, gamified exercise platform for rehabilitation clinicians and their patients (like a AI personal rehab assistant). It turns prescribed therapeutic exercises into interactive games that patients can play on any device with a front-facing camera. Our system uses pose detection and an AI agent to guide movement in real time, giving feedback, encouragement, and reminders to improve adherence and recovery.
What we made today
Today we have made an AI powered platform that delivers patient medical data in a clear and concise dashboard which minimizes time wasted and scheduling conflicts.
Parses patient medical records which can be 200,000+ lines each.
Provides AI driven summaries of medical records.
Assists with scheduling priority of 25+ different patients each day.
How we built it
Using a mock medical industry compliant dataset from Synthea we built a database of patient data which is retrieved and analyzed with Anthropic MCP calls to deliver summaries to a clinician-friendly dashboard. We also used a mock dataset which contains a record of patient history using our gamified physical therapy tools.
Challenges we ran into
Extracting relevant information from extremely detailed and large datasets. Learning and implementing the MCP protocol in a short amount of time. Creating relevant prompts and toolsets which do not waste API calls. Delivering the information in a clear and concise way. rate limiting.
Accomplishments that we're proud of
Successful integration of medical record database retrieval with added AI analysis. A clean and sleek UI for clinicians to easily access data. Building something relevant to our work that we will use going forward. Building familiarity with real world (synthesized, life-like, not real) medical data.
What we learned
How to use MCP. How to integrate AI into large dataset analysis.
Built With
- anthropic
- mcp
- python
Log in or sign up for Devpost to join the conversation.