Accés ràpid intranet

Més informació...

a a a

Deim Seminar


Quantifying Genomic Privacy in Genetic Test Results


Dr Hiromi Arai

Professor/a organitzador/a

Josep Domingo-Ferrer


Universitat de Tquio


23-11-2016 12:00


The wide spread of personal genomic data use raises privacy concerns. We deal with privacy quantification for genetic tests, one of the most popular applications of the personal genome. Disease risks, which are the result of the genetic tests, are calculated on the bases of the users genetic information. We investigate privacy in disease risks in a scenario where an adversary infers a users genome by reverse-engineering from the users disease risks. We developed a computationally efficient method to evaluate this privacy using integer programming techniques. The efficiency and scalability of our method are shown experimentally by using simulated data. We evaluated privacy of a users genome in accordance with several privacy metrics by using real-world datasets. We found that a considerable amount of personal genetic information was inferred from disease risks. Furthermore, we found that a risk factor for cancer can be inferred from non-severe disease risks.


Laboratori 231