Bridging the Gap to Universal Health Information Exchange with Community-Driven Terminology Mediation

The most critical obstacle to the seamless exchange of healthcare information is the translation from the proprietary medical ontologies of private providers to a universal reference language (i.e., the 3M Health Data Dictionary). Stottler Henke proposes to develop a new system, called Bridge, which will directly address this challenge through three primary techniques. First, we employ a novel approach to ontology mapping in which a community of domain and subdomain experts can work together to create more comprehensive and precise mappings with benefits including: higher dynamicity and up-to-dateness to changes over time, a reduction in errors and incomplete mappings through consensus reaching, and avoiding placing too high of a burden on any individual user. Second, we will enhance mapping accuracy by pairing these community contributions with a"committee"of complementary automated ontology mapping systems. Finally, we will incorporate an active learning module to recognize when ambiguity resolution by a human-in-the-loop can have a cascading effect throughout the automated mapping process and thereby prioritize user operations that can profitably constrain/direct the process with little effort. Phase I development of a limited prototype will provide a solid foundation for the complete implementation of Bridge in Phase II and its eventual commercialization.