POSTER: PriReMat: A Distributed Tool for Privacy Preserving Record Linking in Healthcare

Abstract

Medical institutions must comply with various federal and state policies when they share sensitive medical data with others. Traditionally, such sharing is performed by sanitizing the identifying information from individual records. However, such sanitization removes the ability to later link the records belonging to the same patient across multiple institutions which is essential for medical cohort discovery. Currently, human honest brokers assume stewardship of non sanitized data and manually facilitate such cohort discovery. However, this is slow and prone to error, not to mention that any compromise of the honest broker breaks the system. In this work, we describe PriReMat, a toolset that we have developed for privacy preserving record linkage. The underlying protocol is based on strong security primitives that we had presented earlier. This work describes the distributed implementation over untrusted machines and networks.