ARM-Based Privacy Preserving for Medical Data Publishing

摘要

The increasing use of electronic medical records (EMR) makes the medical data mining becomes a hot topic. Consequently, medical privacy invasion attracts people's attention. Among these, we are particularly interested in the privacy preserving for association rule mining (ARM). In this paper, we improve the traditional reconstruction-based privacy preserving data mining (PPDM) and propose a new architecture for medical data publishing with privacy preserving, and we present a sanitization algorithm for the sensitive rules hiding. In this architecture, the sensitive rules are strictly controlled as well as the side effects are minimized. And finally we performed an experiment to evaluate the proposed architecture.