Connectivity and Semantic Patterns in Automatically Generated SNOMED-CT Subsets

Abstract

The importance of enabling a consistent development and use of modern clinical terminologies is nowadays unquestionable by the health informatics community. SNOMED-CT is one of such terminologies whose adoption is fostered by the IHTSDO with a worldwide scope. However, the large scale of SNOMED-CT is a major barrier to its progress while there is evidence that only a small fraction of its content is being used. This paper proposes a series of graph-based methods to analyze and evaluate the fitness and expressiveness of automatically generated SNOMED-CT subsets vis-à-vis the clinical context they were generated for. Such evaluation methods are then applied to the subsets obtained from previous experiments that used clinical guidelines’ glossaries as seeds for the automatic generation of subsets. Current research provides new means to judge if clinical applications of SNOMED-CT can go beyond the traditional uses of a controlled vocabulary or not.