A two-member team from Vanderbilt has won the 2016 iDASH Healthcare Privacy Protection Challenge, a competition open to international participants and devoted this year to privacy protection for genome analysis in a cloud computing environment.
The competition’s two other winning teams were in large part composed of members from industry giants Microsoft and IBM.
Zhiyu Wan
The Vanderbilt team included computer science graduate student Zhiyu Wan and Bradley Malin, Ph.D., associate professor of Biomedical Informatics and Computer Science. Malin directs the Health Data Privacy Lab and is co-director of the Center for Genetic Privacy and Identity in Community Settings (GetPreCiSe). Wan, who is currently working toward his doctorate, is a member of both of these research groups.
Wan and Malin took top honors in track one of the competition, in which teams designed algorithms to thwart re-identification of individuals as de-identified genomic data are made available through the Beacon Network. Their winning algorithm proved best at preserving the scientific utility of a genomic data set while thwarting a privacy attack.
Bradley Malin, Ph.D.
The Beacon Network, a program of the Global Alliance for Genomics and Health, was launched in 2014 to test the willingness of international sites to share genetic data in a specific technical context. The network’s so-called beacons are web servers that allow users to search an institution’s databases, in a strictly limited way, to determine whether they contain genetic variants of interest; beacons respond to queries of the form “Do you have a genome that has a specific genetic variant at a specific genomic position.”
In 2015, researchers at Stanford University found some vulnerabilities in the Beacon Network approach to genetic data sharing. Among their findings: in a beacon with 1,000 individuals, an attacker who had already gained access to an individual’s genome sequence could, with just 5,000 queries, determine whether the individual was included in the beacon. (If the beacon were expressly devoted to data from people with a particular disease, the attacker would have gained new information about the individual.) Track one of the competition called for algorithms to correct these vulnerabilities.
A six-member team from IBM, Cornell University and Bar-Ilan University won track two of the competition, which involved privacy-preserving searches of cancer patient data across organizations.
A seven-member team from Microsoft Research won track three of the competition, which involved testing for genetic diseases on encrypted genomes.
The winning teams were announced Nov. 11 in Chicago at the 2016 iDASH Privacy and Security Workshop.
iDASH, The National Center for Integrating Data for Analysis, Anonymization and Sharing, is funded by the National Institutes of Health under its program of National Centers for Biomedical Computing, and is based at the University of California at San Diego.
Media Inquiries:
Paul Govern, (615) 343-9654
paul.govern@Vanderbilt.Edu