A Comprehensive Survey on Federated Learning and its Applications in Health Care

With the advancements in the Internet of Things (IoT), data collection has become easier and quicker than ever. Data collected from various devices and organizations leads to the requirement of a special kind of processing. Traditional Artificial Intelligence (AI) and Machine Learning (ML) algorithms might not be suitable to process these big data. Apart from computational and storage concerns, preserving personal data privacy is another crucial aspect. Federated Learning (FL) is the decentralized approach that trains the models locally on the device. Only updated parameters are communicated with the global server. This will not only get the benefit of computing power of participating edge devices but also provide inherent data privacy. This paper provides a survey on state-of-the-art work done on various types of cancers, COVID-19, and medical imaging in recent times. The studies have shown improvements in prediction accuracy with a federated model trained on multiple institute data over the traditional model trained on single institute data. The article also highlights the open research issues and the recent developments in the field.