Privacy-preserving disease detection techniques in e-healthcare systems

• Main Author: Wang, Guoming
• Other Authors: Guan Yong Liang
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/138085
• Institution: Nanyang Technological University

With the increasing popularity of pervasive devices such as smartphones, Body Sensor Network(BSN), Internet-of-Things devices and cloud computing, mobile e-healthcare has become a research trend in recent years. Disease detection using big data analytics techniques is a popular e-healthcare research focus. However, big-data disease detection still faces many challenges on privacy of users' sensitive personal information, confidentiality of health service provider's diagnosis model, accuracy of diagnosis, efficiency of query, etc. In this thesis, we aim to improve the security and privacy performance of the e-healthcare systems, while achieving efficient disease detection. Firstly, we propose an efficient privacy-preserving pre-clinical guidance service scheme (PGuide) to provide on-the-go medical guidance service while preserving user privacy. Using the PGuide scheme, users can personally conduct privacy-preserving pre-clinical diagnosis based on their health profiles and obtain recommendation from trusted sources (e.g. hospitals and medical service providers) based on the diagnosis. In addition, the information transmitted to the hospitals and other medical service providers to calculate the disease risk use a disease prediction model in a privacy-preserving way. Secondly, we propose an efficient privacy-preserving health query scheme over outsourced cloud (HeOC). In this scheme, the user shortlists the possible disease on his own using self-collected health sensor data first. Then the user can query the disease severity accurately from the filtered result in the cloud. To reduce the query latency, we propose a novel sensor anomaly detection technique (SADS) for detecting high-risk disease. In the SADS technique, the health service provider outsources an encrypted health tree to the cloud. Authenticated users send encrypted physiological data to the cloud to detect high-risk disease without disclosing his/her sensitive personal health data. Then, with the oblivious pseudorandom function protocol (OPRF), the user retrieves the diagnosis result accurately. Thirdly, we propose an efficient and privacy-preserving priority classification scheme (PPC), for classifying encrypted patient data at the Wireless Body Area Network gateway (WBAN-gateway) in a remote e-healthcare system. Specifically, to reduce the system latency, we design a non-interactive privacy-preserving priority classification algorithm, which enables the WBAN-gateway to perform privacy-preserving priority classification on the received users’ medical packets by itself and relay these packets to the healthcare center according to their criticality, i.e. more critical cases get reported first.