This wiki has undergone a migration to Confluence found Here

Post Coordinated Terminology Notes

From HL7Wiki
Jump to navigation Jump to search

Return to Clinical Decision Support Work Group

Return to Project Page

Clinical information models derive their core semantics through their associations with medical ontologies. These associations are often referred to as terminology bindings and can exist at two levels: (1) at the predicate level and (2) at the predicate range level. Yet, models and terminologies are often developed in isolation of one-another and this can often lead to gaps in terminologies, incongruences between these types of models, and ambiguity in resulting clinical models. In order to avoid such incongruences, recent efforts such as the Clinical Information Model Initiative have aimed to better harmonize their modeling activities with those of SNOMED CT, LOINC, and RxNorm.

Unfortunately, terminology modeling efforts have also occurred within individual silos leading to duplication of concepts across terminologies, different underlying concept models and technological approaches, and incomplete coverage of specific areas in each of the these terminologies. In order to support the needs of the clinical modeling community, better alignment across terminologies is required if these efforts are to scale. This requires a better understanding of the scope of use of each terminology in clinical models and a common underlying approach to the development of complementary concept models to support terminology expressions that can span beyond individual terminologies. For instance, a medication administration procedure may make use of the SNOMED CT Procedure concept model and an RxNorm Medication Product concept model to represent the direct substance administered to the patient. The RxNorm terminology may also add the capability to group medications by class in a manner similar to the DL-based hierarchies supported in SNOMED CT or work with IHTSDO to achieve this aim in a complementary fashion.

At this time, SNOMED CT, through its use of EL-based concept models, has the most computable concept model and expression language. While many of the technologies developed for SNOMED CT could easily be applicable to other terminologies, they are proprietary to IHTSDO and their use limited to SNOMED CT. Thus, in order to support cross-terminology post-coordinated expressions, an investigation of how to generalize such technologies across standard biomedical ontologies could prove to be of great value to clinical modelers and to medical interoperability. [Note: Reasoning over terminologies - Conceptual model, etc - Also account for expressions]