Federated learning for privacy-preserving iot data aggregation in smart healthcare applications

Internet of Things (IoT) gadgets have transformed patient monitoring and individualised treatment through their incorporation into state-of-the-art healthcare systems. Concerns regarding data privacy and security are heightened by the decentralised nature of IoT data collection, as well as by the increasingly inadequate state of traditional centralised machine learning approaches that aggregate sensitive health data on central servers in response to strict privacy regulations such as GDPR and HIPAA. It examined the designs for safe data aggregation techniques like homomorphic encryption and secure multiparty computation, as well as differential privacy, which are safe data aggregation techniques. Also, we talk about incorporating edge and fog computing to improve FL efficiency and scalability within resource-constrained IoT devices. An approach that has some promise is federated learning (FL), which allows distributed devices to train models jointly without explicitly exchanging data. In order to collect data from the Internet of Things (IoT) while safeguarding patient privacy, this article focuses on how smart healthcare settings might utilise FL. The vision overhead, data diversity, and adversarial attacks are critical issues the research tackles that the vision overhead, data diversity, and adversarial attacks are essential issues the research tackles and offers solutions for. By reviewing past research and real-world instances, the study highlights how deeply transformative FL can be for the healthcare field, creating secure and compliant intelligent ecosystems. The results indicated the need for FL IoT frameworks to attain asymmetric efficiency targets focused on privacy, while the security and the obfuscation components are highly balanced and robust to maximize privacy.

Перевод пока недоступен