A blockchain-based federated learning mechanism for privacy preservation of healthcare IoT data.

Autor: Moulahi W; Faculty of sciences and Techniques of Sidi Bouzid, University of Kairouan, Tunisia; REsearch Groups in Intelligent Machines (LR11ES48), Tunisia., Jdey I; Faculty of sciences and Techniques of Sidi Bouzid, University of Kairouan, Tunisia; REsearch Groups in Intelligent Machines (LR11ES48), Tunisia. Electronic address: imen.jdey@fstsbz.u-kairouan.tn., Moulahi T; Department of Information Technology, College of Computer, Qassim University, Kingdom of Saudi Arabia., Alawida M; Department of Computer Sciences and Information Technology, Abu Dhabi University, 59911, Abu Dhabi, United Arab Emirates., Alabdulatif A; Department of Computer science, College of Computer, Qassim University, Buraydah, Kingdom of Saudi Arabia.
Jazyk: angličtina
Zdroj: Computers in biology and medicine [Comput Biol Med] 2023 Dec; Vol. 167, pp. 107630. Date of Electronic Publication: 2023 Oct 31.
DOI: 10.1016/j.compbiomed.2023.107630
Abstrakt: The Corona virus outbreak sped up the process of digitalizing healthcare. The ubiquity of IoT devices in healthcare has thrust the Healthcare Internet of Things (HIoT) to the forefront as a viable answer to the shortage of healthcare professionals. However, the medical field's ability to utilize this technology may be constrained by rules governing the sharing of data and privacy issues. Furthermore, endangering human life is what happens when a medical machine learning system is tricked or hacked. As a result, robust protections against cyberattacks are essential in the medical sector. This research uses two technologies, namely federated learning and blockchain, to solve these problems. The ultimate goal is to construct a trusted federated learning system on the blockchain that can predict people who are at risk for developing diabetes. The study's findings were deemed satisfactory as it achieved a multilayer perceptron accuracy of 97.11% and an average federated learning accuracy of 93.95%.
Competing Interests: Declaration of competing interest We have no conflicts of interest to disclose.
(Copyright © 2023 Elsevier Ltd. All rights reserved.)
Databáze: MEDLINE
Externí odkaz: