Summary
Objectives:
SNOMED® CT is emerging as a reference terminology for the entire health care process. It claims to be founded on logic-based modelling principles. In this article, we analyze a special encoding scheme for diseases and procedures in SNOMED® CT, the so-called relationship groups (RGs), which had been devised to avoid ambiguities in definitions.
Methods:
We reformulate SNOMED® CT’s relationship groups in the format of description logics in order to check whether RGs serve the needs they were designed for.
Results:
We show that a considerable proportion of relationship groups represent hidden mereological relations. We also report discrepancies encountered between the defined semantics of many SNOMED® CT terms and their intuitive meaning, as well as inconsistencies detected between the definition of various complex composed terms and the definition of their top-level parents.
Conclusions:
We formulate recommendations for improving SNOMED® CT by replacing most occurrences of relation groups by formally more adequate “part-of” relations.
Keywords
SNOMED CT - knowledge representation - logic
