PROJECT TITLE :

Adoption of Federated Learning for Healthcare Informatics Emerging Applications and Future Directions

ABSTRACT:

The patients' quality of life (QoL) has significantly improved thanks to the intelligent healthcare system, which allows for remote stakeholders to perform analysis on patients' medical records. In order to train artificial intelligence (AI) models in a manner that is both efficient and effective, it is necessary to exchange a large amount of data using an open communication channel, such as the Internet. This is necessary for disease prediction. The open nature of communication channels creates a significant risk to the confidentiality of collected data and interferes with the process of training models using information stored on centralized servers. An emerging idea known as federated learning (FL), which can be used as a potential solution to this problem, can help. It does the training at the client nodes and then aggregates the results of those training sessions in order to train the global model. The idea of local training helps to maintain the patient's right to privacy, as well as their confidentiality and the integrity of their data, all of which are important contributions to the process of effective training. The application of FL in the field of healthcare has a number of potential benefits, but it has not yet been investigated to the full extent that it possesses. Existing surveys concentrated their attention primarily on the function of FL in a variety of applications, but there has not been a survey that is either detailed or comprehensive concerning FL in healthcare informatics (HI). We present a comparative analysis of recent surveys in relation to the survey that is being proposed. We proposed a FL-based layered healthcare informatics architecture along with a case study on FL-based electronic health records in order to improve the quality of life for patients and the privacy of their medical records (FL-EHR). In this paper, we discuss the developing FL models and present the statistical and security challenges associated with the adoption of FL in medical settings. As a result, this review offers insights that can be helpful for researchers in academic settings as well as healthcare practitioners who are investigating the application of FL in HI environments.