CRQMJul 25, 2016

Realistic DNA De-anonymization using Phenotypic Prediction

arXiv:1607.07501v1
Originality Synthesis-oriented
AI Analysis

This work addresses privacy risks in genomics for individuals and organizations handling DNA data, but appears incremental as it builds on existing approaches.

The paper tackles the problem of linking individuals to DNA sequences through phenotypic prediction, improving upon current methods and proposing defenses against such attacks.

There are a number of vectors for attack when trying to link an individual to a certain DNA sequence. Phenotypic prediction is one such vector; linking DNA to an individual based on their traits. Current approaches are not overly effective, due to a number of real world considerations. This report will improve upon current phenotypic prediction, and suggest a number of methods for defending against such an attack.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes