Privacy-Preserving Medical Reports Publishing for Cluster Analysis

Health data mining is an emerging research direction. High-quality health data mining results rely on having access to high-quality patient information. Yet, releasing patient-specific medical reports may potentially reveal sensitive information of individual patients. In this paper, we study the problem of anonymizing medical reports and present a solution to anonymize a collection of medical reports while preserving the information utility of the medical reports for the purpose of cluster analysis. Experimental results on a collection of real-life medical reports suggest that our proposed approach can effectively preserve both information utility and privacy protection in medical reports.