Federated Learning for Healthcare Privacy-Preserved Artificial Intelligence in Distributed Systems

Abstract

Integrating the Internet of Things (IoT) is prominently embraced in healthcare and is referred to as the Medical IoT
(MIoT). MIoT is transforming healthcare by offering numerous advantages for patients and healthcare professionals.
The utilization of MloT is rapidly increasing, producing substantial volumes of IoT data that necessitate thorough
analysis to derive significant insights. This has resulted in the implementation of Artificial Intelligence (AI) methods,
including Machine Learning (ML) and Deep Learning (DL) algorithms, to comprehend the significance of the
underlying health information, with the process of learning typically occurring in cloud or healthcare systems. The
rapid proliferation of IoT sensors and the extensive distribution of private MloT data sets render centralized learning
AI systems increasingly challenging to implement for these jobs. In this context, Federated Learning (FL) is becoming
recognized as a viable device learning approach without transferring sensitive and confidential information to a central
server. The terminal gear and the central server in FL only exchange learning model upgrades to maintain the
confidentiality of sensitive data. Despite its emergence as a viable research domain, no contemporary studies have
been undertaken. This paper synthesizes recent work and advancements in FL to enhance FL-driven MIoT medical
applications and services. This research enables stakeholders in both academia and business to comprehend the
advantages of the most sophisticated privacy-preserving MloT platforms utilizing FL.