Med Mirror
Inspiration
People only go to doctors when something’s wrong — mainly because access to quality care is difficult. Med Mirror aims to fix that by giving people a digital health twin that can proactively visit AI healthcare specialists and help improve their health before problems appear.
What it does
Med Mirror is a digital twin system for your health.
- Visualizes how you’ll look in 10 years based on your health data and lifestyle.
- Lets your digital twin visit AI doctor specialists designed to optimize your health.
- Provides personalized, actionable advice to help you make smarter health decisions today.
How we built it
- Used CrewAI to build the user’s digital health twin.
- Used CrewAI again to build agentic AI doctor specialists that the twin can visit.
- Used Exa to power the medical accuracy of our AI doctors.
- Used Fly to quickly host both backend and frontend.
Challenges we ran into
- Building useful AI doctor agents that deliver medically accurate guidance.
- Managing agent communication without unnecessary complexity.
- Designing meaningful A2A interactions for health applications.
Accomplishments that we're proud of
- Built a working prototype of both the digital health twin and AI doctor specialists.
- Created an early foundation for A2A interactions between digital patients and doctors.
- Integrated real health data to drive predictions and personalized advice.
What we learned
- A2A is useful for interfacing with agentic systems we don’t control.
- CrewAI is great for building agentic systems we do control.
- Visual health feedback (like predicted aging) strongly engages users.
What’s next for Med Mirror
- Enable full A2A capability so the digital patient twin can autonomously visit various AI doctors.
- Build an MCP server to pull real health data from EHRs and personal health devices.
Built With
- crewai
- exa
- fly
- weave
- weights&biases
Log in or sign up for Devpost to join the conversation.