Towards Privacy-Preserving Forensic Analysis for Time-Series Medical Data
Xiaoning Liu, Xingliang Yuan, Joseph K. Liu · 2018
Electronic medical record (EMR) forensics is at the forefront of both academia and industry, and has dominated increasingly important role in the fast revolutionized digital forensics area. Upon the severe financial loss and user privacy revealing caused by data breaches, protecting the forensic medical records only being mined by authorized investigators and data confidentiality is deemed essential. Standard encryption technique can ensure the end-to-end data security, yet restricting the functionality in forensic analyzing. How to proceed similarity match over forensic physiological data in a private manner is intrinsically challenging, because the natural properties of such medical data are high-dimensional and times series related. In this paper, we propose a secure framework to proceed similarity match over encrypted physiological time-series data. Our framework resorts to an advanced similarity search algorithm, aka stratified locality-sensitive hashing (SLSH) to assist an authorized forensic investigator to have in-depth understanding of physiological data with multiple perspectives. In addition, our framework adopts a scalable encrypted index construction which provides provable security guarantees. Finally, we give a discussion of our future work based on this framework. As a generic and scalable framework, our design can be easily extended to secure update and parallel processing.