A Novel AI Approach for Privacy Preserving In Biomedical Using Federated Learning

One of the pivotal applications of artificial intel-ligence (AI) in the healthcare sector is to reduce dependency on human physicians, or at least diminish the frequency of their necessity. While AI cannot yet entirely replace doctors, it functions as an indispensable tool that significantly enhances their capabilities. This research summarizes various technologies that facilitate the integration of AI into healthcare applications. In particular, the development of effective AI systems necessitates the accumulation of vast datasets. Centralizing this data is imperative, which requires establishing a robust platform capable of storing and providing seamless access to this data for future use. Federated learning, a type of distributed and decentralized machine learning, is particularly useful when users cannot or do not wish to share their data with a central server due t o privacy and security concerns. This paper also summarizes general solutions to the statistical, system, and privacy challenges in federated learning, highlighting some results of machine learning techniques applied to healthcare. Ultimately, this paper can serve as a foundation for guiding future research in AI based on federated learning for healthcare challenges.