Synapse Med-Diagnostics
A HIPAA-compliant portal that helps radiologists spot unusual MRI patterns. Models learn across hospitals without patient data leaving them.
- Client
- A network of three hospitals
- Our role
- AI models, secure portal, compliance
- Year
- 2025
- Built with
- Next.js, Federated learning, Zero-trust encryption
The problem
Each hospital had too few rare cases to train a useful model alone, and sharing scans between them was not allowed.
What we built
We used federated learning so the model travels to the data instead of the other way round, behind zero-trust encryption and a portal radiologists actually enjoy using.
What changed
Radiologists get a second opinion in about a second per scan, and no patient record has ever left its hospital.
- Detection accuracy
- 99.4%
- Per scan
- 1.2 s
- Patient records shared
- 0
Field Notes