Federated Learning Framework Now Live in Two Partner Hospital Networks

The Data & Knowledge Systems Lab's federated learning framework, developed over three years under Professor Priya Raghunathan, is now running in production at two partner hospital research networks. The system allows the institutions to train shared diagnostic models without any patient data leaving its original source. The project began as a response to a specific limitation: promising diagnostic models that couldn't be trained at scale because the hospitals that held the most relevant data weren't able to share it directly.