Federated Learning Applications in Healthcare

Over the last decade, Federated learning has gained rapid attention in its application to data privacy and security in the healthcare sector. Traditional machine learning models often rely on centralized data aggregation, which poses significant risks related to patient privacy and data security. Federated learning offers a different approach that enables the training of machine learning models across multiple institutions without the need to share sensitive patient data. This approach not only preserves privacy but also enhances the scalability, adaptability, and overall utility of machine learning in healthcare applications and scenarios. 

However, practical clinical deployments reveal major gaps between theory and practice. Healthcare data are highly heterogeneous, institutional resources and network quality vary widely, and threat models are stronger than typical FL assumptions. Further, privacy guarantees often require complementary mechanisms such as differential privacy, secure aggregation or cryptography, that trade off accuracy, latency, or computational cost. 

This survey aims to provide a review of FL in healthcare, analyzing representative case studies that demonstrate the practical viability of Federated Learning in healthcare settings. For each, we extract empirical lessons about convergence, accuracy of communication trade-offs, privacy measures, and deployment feasibility. The study takes a deliberately skeptical stance, exposing common assumptions, discussing failure modes, and highlighting practical mitigations. The scope is to provide an insight of the implementation of FL in healthcare, encompassing both foundational review articles and specific case studies that demonstrate practical implementations. To guide this comprehensive review, the following research questions will be addressed:

• What are the different implementations of Federated Learning in healthcare?
• What challenges are associated with implementing Federated Learning in healthcare?
• What are the potential future developments and research opportunities in Federated Learning?