Exploring semantics in clinical data interoperability

Abstract

The increasing amount of digital clinical information has prompted research in interoperating across numerous clinical data sources. Most solutions to this problem follow one of two main directions: (a) adoption of Electronic Health Records (EHR) standards, or (b) structuring medical knowledge via Knowledge Organizations Systems (KOS). Related research sometimes addresses the combination of the two directions but does not explore the knowledge of KOS, which are just used to define and disambiguate concepts. This paper discusses the solution for clinical data interoperability that we designed and implemented. It is a two step process - (a) we provide initial integration via mediators to provide mappings across heterogeneous sources, and (b) KOS to extend navigation possibilities across data sources. This paper is centered in the second part, discussing the challenges of semantic query expansion for clinical data analysis. We illustrate our solution through a real case study from one of Brazil’s largest hospital complexes.