Data privacy limited collaborative modeling
Sensitive patient data prevented institutions from training shared predictive models.
We enabled privacy-preserving collaboration
Federated learning allowed joint model training without sharing raw patient data.
Federated ML with NVIDIA FLARE
Models were trained using PyTorch across structured, time-series, and unstructured clinical data within a federated setup.
Promising early cancer risk prediction performance
Initial results demonstrated strong predictive potential while preserving patient privacy.
