Re-identification of Clinical Data Through Diagnosis Information

In this chapter, we present an attack that can associate patients with their diagnosis and genomic information. The attack involves linking the published data with external, identified datasets, based on diagnosis codes. After motivating the need to prevent the attack, we discuss the type of datasets that are involved in the attack, in Sect. 3.2. Then, a measure that quantifies the susceptibility of a dataset to the attack, as well as a study of the feasibility of the attack in an Electronic Medical Record (EMR) data publishing scenario, are presented in Sect. 3.3. Last, a set of measures that capture the utility loss that sharing the published data in a way that prevents the attack is discussed in Sect. 3.4.