基于双层建模的知识表达方法在医学知识库构建中的应用

A Knowledge Representation Method Based on Two-Layer Modeling for Constructing Medical Knowledge Base

医学知识库是临床决策支持系统(CDSS)的重要组成部分。目前部署在医院内部的CDSS通常针对特定的需求(如合理用药、辅助诊断),其知识库所涵盖的知识种类比较单一。构建一个整合多种不同医学知识种类的协同知识库,能够使得不同类型的知识内容作为一个有机整体提供更加全面和有效的决策支持应用。提出一种基于双层建模的知识表达方法,首先分析和归纳临床信息化环境下影响诊疗决策的各类医学知识,如诊断知识、治疗知识、药学知识等,并对它们之间的协同关系进行定义,形成认知模型,然后针对特定的知识推理和计算的需求,为每一种知识类型选取具体的知识表达形式,形成最终的计算模型。基于该双层建模方法所研发的临床决策支持系统在三甲医院已运行4年多,实践表明该系统有效解决不同知识类型的集成和推理的需要,为构建整合和全面的知识应用奠定基础。

Medical knowledge base is an important asset for clinical decision support system （CDSS）. Current CDSSs deployed in hospitals are mostly targeted at specific requirements （e. g. rational drug use, computer- aided diagnosis） , and their knowledge bases usually contain one specific type of knowledge. A comprehensive knowledge base that can coordinate different types of medical knowledge is essential for building integral and effective decision support applications. This paper proposes a knowledge representation method based on two- layer modeling. First, an epistemological model is constructed to cover various diagnostic and therapeutic knowledge types used in clinical settings, as well as their coordinative relationships. Then, based on specific reasoning and computation requirements, concrete representation formulisms are assigned to each knowledge type to form the final computational model. A clinical decision support system based on this two-layer modeling method has been running for more than 4 years in a Class III hospital. Practices show that the system effectively solved the need of integrating and reasoning of different knowledge types, and has established a foundation for building integral and comprehensive knowledge applications.