Why Methods for Genomic Data Privacy Fail and How We Can Fix It

The increasing integration of patient-specific genomic data into clinical practice and research raises serious privacy concerns. Various privacy enhancing techniques that “de-identify” the data through the removal of explicit identifiers, such as name or Social Security number, have been proposed and deployed. While advocates of these systems have the best of intentions, they are fundamentally flawed, due to a lack of formal modeling and proofs of privacy. These systems fail to account for information that can be inferred from genomic data, as well as, the environment into which data is shared. This research addresses the extent to which these systems are susceptible to computational “re-identification” attacks. The attacks we employ exploit residual information in genomic data and relates it to explicit identity. Though susceptibility varies, each of the protection methods studied is deficient in their protection against re-identification. Our findings stress the need for genomic data privacy protection methods that allow for provable guarantees of privacy.