Genomic Privacy Metrics: A Systematic Comparison

The human genome uniquely identifies, and contains highly sensitive information about, individuals. This creates a high potential for misuse of genomic data (e.g., Genetic discrimination). This paper investigates how genomic privacy can be measured in scenarios where an adversary aims to infer a person's genome by constructing probability distributions on the values of genetic variations. Specifically, we investigate 22 privacy metrics using adversaries of different strengths, and uncover problems with several metrics that have previously been used for genomic privacy. We then give suggestions on metric selection, and illustrate the process with a case study on Alzheimer's disease.