A risk management framework for health care data anonymization

Abstract

To facilitate many important tasks ranging from medical research to personalized medicine, micro datasets that contain sensitive patient information need to be shared. To address this important issue, there has been considerable work done on anonymization techniques that try to protect privacy. At the same time, under various assumptions about the background knowledge available to adversary, it is shown that most of the existing microdata anonymization techniques could be attacked. In this paper, instead we suggest a novel risk management framework that could be used to analyze risk in sharing anonymized data.