A STUDY ON DEVELOPMENT OF LARGE-SCALE MEDICAL ONTOLOGY FOR CLINICAL USE

为了开发用于临床的大规模医学本体，我们在研究期间进行了以下四个课题。1)医学词典中临床术语间关系的自动提取与分类2)人体要素间关系的自动提取3)诊断报告中重要临床发现的自动提取4)大规模医学本体数据库结构的考虑在完成上述工作后，我们使用本体编辑器“HOZO”开发了一个顶级本体框架，用于内科、外科、临床发现、外科手术和解剖学。例如，一般疾病概念被分为可以用疾病的病因来表示的正式疾病和只能用临床表现和症状列表来表示的证候。对于解剖结构，引入实体来表示解剖元素，并表示实体与相邻结构之间的空间关系。该方法可以表示所有的解剖实体。我们验证了3000种疾病可以用这个顶层本体框架来表示。

For the purpose of developing large-scale medical ontology for clinical use, we conducted the following four subjects in the research period.1) Automatic extraction and classification of relationship among clinical terminology from medical dictionary2) Automatic extraction of relationship among human body elements3) Automatic extraction of important clinical findings from diagnostic reports4) Consideration about database structure for large-scale medical ontologyAfter the above work, we developed a top-level ontology framework, using ontology editor "HOZO", in the domain of internal medicine, surgery, clinical findings, surgical procedure, and anatomy. For example, general disease concept was classified into formal disease that can be represented by the etiology of the disease, syndrome that can be represented only by the list of clinical findings and symptoms. For anatomical structures, entity for representing an anatomical elements was introduced and the spatial relationship between the entity and the adjacent structures was represented. All the anatomical entities could be represented by this method. We verified that three thousand diseases could be represented by this top-level ontology framework.