Practical Applications of Secure Multi-Party Computation for Public Health and Post-Marketing Drug Surveillance

• Activity Number: 125
• Type: Topic Contributed
• Date/Time: Monday, August 10, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Khaled El Emam and Luk Arbuckle*
• Companies: and CHEO Research Institute

Abstract

Accessing data for analytics purposes can be facilitated if the data is no longer considered personal information. Statistical and computational disclosure control methods can be used for that purpose. Another approach that is more suitable under some use cases is secure multi-party computation. SMC is well suited in the context of surveillance problems because the computations can be defined and optimized and then continuously applied. We will explain how SMC can be applied in health surveillance contexts, with some theoretical and practical results. One project developed a secure linking protocol to link different data sets without sharing personal information. Public health use cases for secure linking include performing HPV vaccination evaluations and Chlamydia testing from general practices. Another example is antimicrobial resistant infection surveillance in long term care homes. This allowed the collection of colonization and infection data without revealing the rates for any of the participating homes. We will also discuss a post-marketing surveillance project where rare drug adverse events are modeled using logistic regression by securely pooling data from multiple sites.