Protecting Privacy in IoMT-based Disease Diagnosis Using an Efficient Approach

Abstract

The profound fusion of the Internet of Things (loT) with the medical sector has resulted in the emergence of the Internet of Medical Things (loMT). Within IoMT, healthcare professionals address patient illnesses by examining data gathered from medical sensors in which their measurements are reported via mobile devices' artificial intelligence (AI)- empowered applications. Hence, the patient's health records are fed into machine-learning (ML) models for disease diagnosis (DD) and prediction. Nevertheless, revealing this sensitive information enables inferring confidential information about patients, thereby, their privacy is violated. To address this privacy issue, the existing works focus on using federated learning (FL)-based approaches to train and obtain an accurate global DD model through collaborative efforts among multiple parties (e.g., healthcare institutions). However, there is a notable gap in research on addressing the privacy violation problem during the future DD process (in the deployment phase) after obtaining the global DD model, which has not been thoroughly investigated yet. Therefore, in this paper, we introduce an innovative privacy-preserving scheme for an accurate disease diagnosis (DD) while emphasizing the preservation of patients' privacy. The proposed design enables patients to encrypt their health data using functional encryption (FE), allowing for DD without revealing or learning patients' data to protect their privacy. Additionally, we develop a hybrid deep learning (DL) model that can yield precise DD. To demonstrate the feasibility of our approach, we assess its performance on real health data records using the Cleveland dataset, from the University of California Irvine (UCI), showing that our scheme can accurately diagnose heart disease while ensuring robustness, privacy preservation, and reasonable overhead.

Topics & Concepts