基于临床知识图谱的慢性肾病辅助决策模型研究

Analysis of Chronic Kidney Disease Decision Support Model Based on Clinical Knowledge Graph

提出了一种基于知识图谱的临床辅助决策系统构建方法。首先,通过建立标准数据模型,对多源数据进行汇聚,建立医学语料库;通过专家路径及医学教科书,参考UMLS的两层结构建立知识图谱的主干;然后通过医学自然语言处理及人工标注机制,对数据进行标准化处理,并通过数据挖掘建立概率模型,为知识图谱补充枝叶;基于知识图谱建立智能问诊路径,解决医生与辅助决策系统的交互问题。通过以上的路径建立起面向基层医生的肾病临床辅助决策系统,实现基层医生对慢性肾病标准化、规范化诊断和转诊,做到慢性肾病早发现、早治疗,起到了较好的效果。

A method of constructing clinical decision-making system based on knowledge graph is proposed.Firstly,medical corpus is built by building standard data model and aggregating multi-source data. Through expert path and medical textbooks,the backbone of knowledge graph is established by referring to the two-layer structure of UMLS.The standard medical language corpus is built by common data model,and at the same time gives the medical language processing tools and human-annotation rules to broaden and modify the knowledge map. Based on this knowledge map,the intelligent inquiry path is established to solve the interaction problem between doctors and assistant decision-making system.Through the above paths,the intelligent CKD decision support model for primary doctors is established,which can guide the primary physicians work standardly and regularly,and guarantee CKD early detection and treatment.