Privacy Technologies for Large Research Databases

Venue

Spectrum Health and Michigan State University, Grand Rapids, MI, September 23, 2005.

Slides from talk

Abstract

"How should researchers and administrators think about privacy when constructing a longitudinal, large-scale person-specific research database across medical, educational, and public service venues?" Constructing such a community-wide database that is minimally invasive to privacy while also linking birth information, school records, public services, and medical data over time is problematical. Issues include: the nature of the entity that would hold and govern the resulting database, its specific research purposes, the fact that the database is assembled from data collected for other purposes, the assortment of regulations (e.g., public health laws, HIPAA, FERPA, state law) that would apply, and the disparity between the beneficiaries (researchers and research dollars) and the subjects of the data (e.g., low income women and children, who receive no direct personal benefit). While the privacy issues are large, complex and diverse, methods based on consent and/or anonymity are crucial to identifying solutions in which the resulting database is provably minimally invasive to privacy while remaining practically useful. This talk will examine these issues in detail, define privacy risks using real-world examples, and explore methods for constructing large-scale databases having provable guarantees of privacy protection while allowing information to be widely shared.

References and Related Links

• Information explosion
• Genomic Privacy
• Genomic System Evaluation
• Trails Learning
• Re-identification, DNA

• Selective Revelation
• Bioterrorism Surveillance
• Homeland Security
• De-identification, Datafly
• k-anonymity
• k-anonymity, Generalization

• Projects at the Data Privacy Lab
• Data Privacy Laboratory