Exploratory evaluation of homomorphic cryptography for confidentiality protection
R21 end date 03/31/2014: National Institutes of Health
(Objective: Confidentiality protection is proving to be a major impediment to public health research. This project will evaluate the feasibility of homomorphic encryption technology that make possible, for the first time ever, analysis without having to decrypt the data. This advance is expected to significantly accelerate research that
involves accessing and/or linking confidential data.)
This project will perform an exploratory evaluation of homomorphic cryptography for the geospatial analysis of confidential health data and will accomplish four aims:
Aim 1: Build a prototype secure multi-party computation platform for the computation of mathematical operations required in geospatial analyses. The platform will implement geospatial analysis protocols on encrypted data such that the identity of individuals cannot be reconstructed or deduced. We will evaluate security of the approach using external expert testing designed to reconstruct identity of individuals. This exploratory project will evaluate the computational performance of the following geospatial operations on encrypted data:
- Spatial weight calculations for residential locations;
- Cluster/hotspot analysis;
- Calculation of rates of late stage diagnosis by race; and
- Calculation of relative and absolute disparities in stage at diagnosis.
These have been selected to be representative of the geospatial computations frequently undertaken in geohealth analyses.
Aim 2: Apply the prototype systems to assess racial disparities in stage at diagnosis for prostate and breast cancers. This will evaluate practical feasibility using previously analyzed data, and will determine whether the results with the not-encrypted data are reproducible.
Aim 3: Formally evaluate the approach and formulate recommendations using an independent working group convened by the North American Association of Central Cancer Registries to include stake-holders including health researchers, IRB Chairs and committee members, experts in confidentiality protection, Directors of disease registries and cryptographers.
Aim 4: Disseminate the recommendations and results of the feasibility analysis though peer-reviewed publications and presentations at scientific meeting. This highly innovative and high-impact project potentially will accelerate human health research, leading to earlier advances in treatment and improvements in our nation’s health.
BioMedware – NAACCR Workshop:
Evaluation of Homomorphic Cryptography for Geospatial Studies with Human Subjects
(Project Summary document –> download)
June 8-9, 2013 in Austin, TX
Geoffrey M. Jacquez 1,2, Khaled El Emam 3, Betsy Kohler 4 and Mike Bykowski 5
1 BioMedware, Ann Arbor, MI
2 Department of Geography, SUNY Buffalo, Buffalo, NY
3 University of Ottawa, Ottawa, ON
4 North American Association of Central Cancer Registries Inc., Springfield, IL
5 Consolidated Solutions and Innovations, Gaithersburg, MD
This workshop was funded as part of the project “National Library of Medicine (NLM) grant “R21 LM011132 Exploratory evaluation of homomorphic cryptography for confidentialityprotection”, Geoff Jacquez Principal Investigator, with co-Investigators Khaled El Emam and Betsy Kohler. We thank workshop participants Francis Boscoe, David O’Brien, Glenn Copeland, Rich Pinder, David Stinchcomb, and Xiao Cheng Wu. Charlie Blackburn organized the workshop, with proceedings recorded by Mike Bykowski.