Recent methodological advances in federated learning for healthcare

Federated learning (FL) promises to solve the challenges of applying machine learning methods within healthcare, such as isolated datasets, ethical, privacy, and logistical concerns with data sharing, and the lack of diversity in single-center datasets. By connecting multiple sites and keeping data at their source, FL can help address these issues. An increasing number of novel FL methodologies have been developed, and there is an urgent need for the consolidation of this knowledge to address the unique challenges of healthcare data. This review, conducted by CMIH Hub researchers, focuses on literature published between 2015 and 2023 that methodologically advances FL for healthcare applications.

Read more about this work in Cell.

Zhang F, Kretuer D, Chen Y. et al. Recent methodological advances in federated learning for healthcare. Cell (2024). https://doi.org/10.1016/j.patter.2024.101006