D&D: A Distributed and Disposable Approach to Privacy Preserving Data Analytics in User-Centric Healthcare

Abstract

Benefiting from the modern information and communication technologies, user centricity has become a clear evolution trend in healthcare. Unfortunately, given the high sensitivity of health data and the uncertainty in user environments, user-centric healthcare systems inevitably suffer from more frequent privacy threats, not to mention that technologies and business of data exploitation have generally outpaced the current privacy regulations and laws. Although there exist well-defined privacy preserving mechanisms, such as Data Encryption, Data Perturbation, and De-identification, they have been considered inadequate to address the diverse privacy challenges in big healthcare data analytics. Our argument is that, before considering any sophisticated mechanism, practitioners should first try to imitate human memory's forgetting process as an intrinsic privacy preserving strategy in user-centric healthcare. Technically, we implement this strategy by changing traditional data analytics routines into a distributed and disposable manner, so as to naturally exclude the data owners' sensitive information. The technical implementation essentially acts as a concrete How-To solution to satisfying a fundamental principle of privacy law, i.e. data minimization. We have initially applied our work to a smart bed project for sleep quality analytics, and received positive feedback on the effectiveness of privacy preservation in suitable homecare scenarios.