Information-Content-based Measures for the Structure of Terminological Systems and for Data recorded using these Systems

Terminological systems such as SNOMED CT play an increasingly important role in contemporary record keeping. This drives the need of assessing the content of these systems, as well as the content of medical records captured using these systems. In this paper, the use of information content as a measure for the structure of terminological systems and the content in medical records is explored. Two complementary information-content-based measures for terminological systems are proposed: the proportion of concepts with zero information content, and the average information content. The measures are applied to the latest releases of SNOMED CT. The measures are useful as an indicator of the overall structure of terminological systems or parts thereof. Furthermore, two measures are described which can provide an estimate for the content of medical records that is captured using a terminological system. Information content is shown to provide a useful basis for assessing the structure of terminological systems and the content of medical records.