Development of a Medical Ontology Library: Analysis of the Clinical Terms in the Medical Records of a COPD Patient / 대한의료정보학회지

OBJECTIVE: The medical records of a patient with chronic obstructive pulmonary disease(COPD) were analyzed to extract medical concepts and their relationships in order to construct a basic medical ontology. METHODS: The medical records included the admission note, vital signs record, doctors' order sheets, progress notes, emergency notes, discharge summary, surgical record, and anesthesia record. RESULTS: A total of 396 concepts, 16 relationships, and 460 connections were created. Fourteen top-level concepts, such as body, sign, and procedure, were found. The most common relationship was 'isA' and the second was 'isPartOf'. All the relationships between the concepts were displayed using the graphic tool GraphViz. CONCLUSION: A pilot ontology on COPD was constructed through a medical record analysis. The asynchronous cooperation using a web interface for the ontology construction was helpful.

How to Deal with Medical Narratives based on Controlled Natural Language Operated by an Ontology / 대한의료정보학회지

OBJECTIVE: Medical narratives entry is a major issue to be solved in developing an electronic medical record system operating in practice, as they are, in large part, described in a free-text format. The issue can be dealt with in three aspects: to improve the reusability by structuring medical narratives, to support clinical pragmatics in medical fields, and to reduce the burden of data entry. With the aspects having in mind, this paper purports to present an ontological method for better way of medical narratives entry. METHODS: We developed an ontology for which medical knowledge is structurally represented. Then we can enter medical narrative texts with commands of the controlled natural language operable on the ontology model. RESULTS: Many theoretical studies on free-text entry were reviewed, based on which an authoring and editing tool for natural language description operable on the ontology model has been developed and tested. The performance of the tool is satisfactory within the limit of the domain models we developed here. CONCLUSION: The results of this paper are contributive for clinicians to make an easy entry of medical narratives as far as the ontology model covers their knowledge domain. It is also expected that the cost in recording medical narratives might be considerably reduced and data quality can be improved.