The current healthcare ecosystem heavily relies on smart services such as disease detection, intelligent drug recommendation, and real-time patient monitoring. In such a scenario, ensuring patient privacy and securing sensitive health data present significant challenges. To address these issues, this article proposes a secure digital twin-enabled ML-based disease detection framework. The proposed framework allows real patients to securely synchronize their real-time health data within a privacy-preserving environment to their corresponding digital twin, which is equipped with ML-based disease detection capabilities. This enables patients to effectively monitor their health vitals and securely identify potential future risks. To ensure security and robustness, the proposed system employs a lattice-based authentication scheme that is resistant to quantum threats. Furthermore, the article conducts experiments on both the authentication mechanism and the ML-based disease detection process to evaluate the robustness of the proposed approach. The results demonstrate that a secure data pipeline from data collection to processing on the digital twin server is essential, as experiments with white-box attacks reveal that models can collapse easily in the absence of proper security mechanisms.
Pandey et al. (Thu,) studied this question.