Which approach BEST describes federated learning in the context of multi-site healthcare data analysis?
-
A
Models are trained locally at each site and only model parameters are shared, not raw patient data
-
B
Patient records are encrypted and transferred to a central server for pooled analysis
-
C
Each site independently publishes summary statistics that are then manually combined by analysts
-
D
A single institution hosts all data under a business associate agreement with every contributing site