PPDDS: A Privacy-Preserving Disease Diagnosis Scheme Based on the Secure Mahalanobis Distance Evaluation Model

The development of Big Data and cloud computing has brought great progress to medical diagnosis and clinical services. However, as disease diagnosis technologies usually use lots of clinical medical data, patients’ privacy becomes increasingly important since the clinical results of medical data are much sensitive. In this article, to deal with the privacy preservation issue of patient’s medical data, we propose a privacy-preserving disease diagnosis scheme that is based on the Mahalanobis distance test, in which the role of the system consisting of query user (QU), aided cloud server (ACS), and classification cloud server (CCS) is to jointly calculate and protect the diagnosis data over the sensitive and outsourced medical data. In the diagnosis model, we outsource the clinical diagnosis process to be dealt with by the CCS, and then, the ACS can reduce the computational cost of the user side. We utilize the homomorphic re-encryption scheme to realize a secure computation over the outsourced medical data, and then, employ a secure multiplication (SM) protocol to implement the privacy-preserving Mahalanobis distance to output the disease diagnosis. Concretely, we provide an extended SM algorithm to solve the problem of multiplication of two encrypted data, and a minimum value comparison algorithm over ciphertext for comparing the encrypted Mahalanobis distance. Finally, we give the experimental performance in real data sheets, and the experimental results indicate that our scheme provides a lower computational cost in practical diagnosis services.