Summary
Objective: To provide a generic approach for developing a domain-specific interface terminology
on SNOMED CT and to apply this approach to the domain of intensive care.
Methods: The process of developing an interface terminology on SNOMED CT can be regarded as
six sequential phases: domain analysis, mapping from the domain concepts to SNOMED
CT concepts, creating the SNOMED CT subset guided by the mapping, extending the subset
with non-covered concepts, constraining the subset by removing irrelevant content,
and deploying the subset in a terminology server.
Results: The APACHE IV classification, a standard in the intensive care with 445 diagnostic
categories, served as the starting point for designing the interface terminology.
The majority (89.2%) of the diagnostic categories from APACHE IV could be mapped to
SNOMED CT concepts and for the remaining concepts a partial match was identified.
The resulting initial set of mapped concepts consisted of 404 SNOMED CT concepts.
This set could be extended to 83,125 concepts if all taxonomic children of these concepts
were included. Also including all concepts that are referred to in the definition
of other concepts lead to a subset of 233,782 concepts. An evaluation of the interface
terminology should reveal what level of detail in the subset is suitable for the intensive
care domain and whether parts need further constraining. In the final phase, the interface
terminology is implemented in the intensive care in a locally developed terminology
server to collect the reasons for intensive care admission.
Conclusions: We provide a structure for the process of identifying a domain-specific interface
terminology on SNOMED CT. We use this approach to design an interface terminology
on SNOMED CT for the intensive care domain. This work is of value for other researchers
who intend to build a domain-specific interface terminology on SNOMED CT.
Keywords
Terminological system - interface terminology - SNOMED CT - subsetting - APACHE IV
