Federated learning improves site performance in multicenter deep learning without data sharing.
Karthik V SarmaStephanie HarmonThomas SanfordHolger R RothZiyue XuJesse TetreaultDaguang XuMona G FloresAlex G RamanRushikesh KulkarniBradford J WoodPeter L ChoykeAlan M PriesterLeonard S MarksSteven S RamanDieter EnzmannBaris TurkbeyWilliam SpeierCorey W ArnoldPublished in: Journal of the American Medical Informatics Association : JAMIA (2021)
Federated learning is an effective methodology that merits further study to enable accelerated development of models across institutions, enabling greater generalizability in clinical use.