Mapping medical records of gastrectomy patients to SNOMED CT

Abstract

The purpose of this study is to explore the ability of SNOMED CT to represent narrative statements of medical records. Narrative medical records of 281 hospitalization days of 36 patients with Gastrectomy were decomposed into single-meaning statements, and these single-meaning statements were combined into unique statements by removing semantically redundant statements. Concepts from the statements describing patients' problems and treatments were mapped to SNOMED CT concepts. A total 4717 single-meaning statements were collected and these single-meaning statements were combined into 858 unique statements. Out of 677 unique statements describing patients' problems and treatments, about 85.5% statements were fully mapped to SNOMED CT. The rest of the statements were partially mapped. This mapping result implies that physicians' narrative medical records can be structured and used for an electronic medical record system.