基于EHR结构模型和DCM的医学术语协同化方法研究

The clinical data can be of great value if they could be shared among computer systems. Therefore, the meaning of data should be represented accurately, explicitly and consistently, especially the medical concepts underpinning the data, as the ways a standard medical terminology system does. However, the majority data generated during care delivering are collected and stored separately by various systems using different medical terminology standards and also local terminologies. Thus, to harmonize those various terms based on their meanings is a useful way to facilitate the comparability and usability of data, while currently there are no such methods available. This study dedicates on developing such method by taking the context of the terms being used in clinical documentations into account. Based upon the standard medical terminologies and related information standard initiatives available internationally, this study makes use of the logical structure of EHR and detailed clinical models, employs ontology development technique and protégé as the tool, meanwhile considering the results from consulting clinical professionals and analyzing medical record data, to develop a meta ontology of clinical terms expressing their clinical context, and then creating a clinical term ontology and repository by constraining and specifying the concepts gradually in depth in the meta ontology. Additionally, this research will apply reasoning capability of OWL-DL to explore the approach of extending the ontology by inferring, and also querying, retrieving and mapping concepts from the ontology to figure out a solution of medical term harmonization.

医疗信息在计算机系统间实现共享才能体现其利用价值。因此，数据含义的表达形式必须准确、清晰和一致，尤其是数据背后医学概念的规范化表达，如医学术语标准。但是，目前大量临床数据由不同信息系统分别采集并分布保存，使用多种标准术语和本地术语。所以，开展术语之间的语义协同对提高数据的可比性和可用性具有重要意义，而目前缺乏面向临床应用的协同化方法。本研究面向医学术语临床应用，在国内外已有标准医学术语系统和相关研究成果的基础上，利用临床记录对象的组合形式和逻辑关联，即EHR文档结构模型和临床信息模型(DCM)，采用protégé本体开发技术和工具，参考临床专家意见和电子病历实例数据，开发反映术语应用语境的临床术语元本体，并对其中的概念逐层约束和特化，形成临床术语本体和术语资源库。同时，利用OWL-DL的推理功能，探索解决本体拓展及概念查询、提取、映射等问题的途径，最终形成面向临床应用的医学术语协同化方法。

本项目聚焦医学术语的规范化表达及不同术语系统之间的协同问题，旨在为提高健康医疗领域数据的计算机可读性和可用性提供方法上的探索。项目计划在国内外已有标准医学术语系统和相关研究成果的基础上，利用临床记录对象的组合形式和逻辑关联，即EHR文档结构模型和临床信息模型DCM，采用protégé本体开发技术和工具，开发反映术语应用语境的临床术语元本体，形成临床术语本体、术语资源库和面向临床应用的医学术语协同化方法。项目实施以来，课题组根据计划书安排，首先全面系统地学习和了解了与本项目相关的国内外医学信息标准研究成果和现状，特别是国际上有影响力的医学信息标准及术语系统，如SNOMED CT，CDA，RIM等；其次，以我国卫生行业标准中已有的数据元和医学术语为基础，对其进行了扩展、整理、分类和建模；第三，参照国际规范的EHR文档结构和DCM，利用protégé开发工具，探索了医学术语本体的构建方法，定义了类及类的层级关系，设置了类的属性、属性的约束和实例，形成了基于语义的网状医学术语资源库，并完成了可视化展示；最后，通过protégé的逻辑推理功能，探索了不同术语资源之间的映射和协同化方法。本项目所进行的研究探索对实现中文背景下医学概念语义的一致性表达具有一定科学价值，研究成果对未来构建中文标准医学术语系统，以及促进不同医学术语系统之间的转换和对照，从而推动我国医学信息化建设、加快健康医疗大数据的分析利用具有重要的应用价值。