Federated Learning for Medicine (FL4M) Consortium
CAMCA collaborates with Rhino Health for a hospital-based federated learning for medicine consortium, “Federated Learning for Medicine (FL4M)”. The consortium consists of seven global healthcare institutions across four continents, targeting enabling healthcare AI development on distributed patient data without moving data from its original healthcare institution. The first algorithm being improved by the consortium is focused on brain aneurysm detection.
The study will focus on detecting undiagnosed brain aneurysms. Developers are training AI models across a federated network of seven global hospitals – using federated learning powered by Rhino Health, so that the patient data never leaves the original hospital site. The ability to run an AI algorithm on globally diverse data sets locally increases the generalizability of the data.
The consortium contains hospitals across a wide range of geographies and patient demographics, caring for over 50 million patients annually. As the institutions span various data privacy regulations and IT infrastructures, federated learning is a highly valuable method that allows the researchers to improve the algorithm’s performance so that it is more generalizable and decreases bias. Early results show differences in algorithm performance at different sites due to differences in patient data, clinical protocols, and scanner types, which highlights the importance of validating and retraining algorithms on diverse data from around the globe. The results have been submitted for peer review.
The FL4M consortium currently includes the following:
Massachusetts General Hospital (in the U.S.): Director of Center for Advanced Medical Computing and Analysis (CAMCA) and Associate Professor at Harvard Medical School, Quanzheng Li, Ph.D.; Instructor at MGH and HMS Dufan Wu, Ph.D.; and Research Fellow in Radiology, Daniel Montes, MD
University of Cambridge School of Medicine (in the UK): Associate Professor Tomasz Matys, MD, Ph.D., MPhil, FRCR, and Professor & Head of Radiology Fiona Gilbert, MD, FRCP, FRCR
Lahey Hospital and Medical Center (in the U.S.): Chair for the Informatics Commission of the American College of Radiology and Chair of Radiology Christoph Wald, MD, Ph.D., MBA, FACR
Assuta Medical Centers (in Israel): Head of Ventures & Innovation and Head of Medical Imaging, Michal Guindy, MD
Dasa S.A. (the largest integrated healthcare network in Brazil): Superintendent of Applied Innovation and AI, Felipe Kitamura, MD, Ph.D., and Head of AI Diagnostic Operations, Paulo Kuriki, MD
National Taiwan University (in Taiwan): Professor of the Institute of Applied Mathematical Sciences and Director of the Medical Data Analytics Laboratory (MeDA Lab) Weichung Wang, Ph.D.
Seoul National University Hospital (in South Korea): Assistant Professor Young-Gon Kim, Ph.D.
For more details about the consortium, please check the news from Rhino Health below: