基于知识图谱的郁病智能辅助辨证知识表示与推理研究

Research on Knowledge Representation and Reasoning of Intelligent Assisted Syndrome Differentiation for Depression Based on Knowledge Graph

以郁病辨证为例,开展基于知识图谱的中医智能辅助辨证知识表示与推理研究,提高中医智能辅助辨证模型的构建效率、辨证模型中辨证知识的可视化程度和辨证推理过程的可解释性。以面向智能辅助辨证的郁病辨证知识获取和医案采集工作的成果为基础,构建郁病智能辅助辨证知识图谱,在知识图谱中表示症状知识和证型知识以及二者之间的关系,结合概率推理进行辨证推理测试和分析。构建了刻画19种证型和147个症状之间关系的郁病智能辅助辨证知识图谱,辨证推理测试获得的初步准确率可达79.17%、按证型分组统计的准确率最高可达100%,可根据郁病智能辅助辨证知识图谱对辨证结果进行初步解释。将知识图谱应用于中医智能辅助辨证知识表示并结合概率推理方法进行辨证推理,有助于提高辨证模型的构建效率和模型中辨证知识的可视化程度。

Taking the syndrome differentiation of depression as an example,this paper carries out research on knowledge representation and reasoning of intelligent assisted syndrome differentiation for Traditional Chinese Medicine(TCM)based on knowledge graph,to improve the construction efficiency of TCM intelligent assisted syndrome differentiation model,the visualization degree of syndrome differentiation knowledge in the syndrome differentiation model,and the interpretability of syndrome differentiation reasoning process.Based on the achievements of knowledge acquisition and medical case collection work for intelligent assisted syndrome differentiation of depression,an intelligent assisted syndrome differentiation knowledge graph for depression is constructed.It represents symptom knowledge,syndrome type knowledge,and the relationship between the two in the knowledge graph,and the probabilistic reasoning is combined to conduct testing and analysis of syndrome differentiation reasoning.A knowledge graph of depression intelligent assisted syndrome differentiation is constructed to depict the relationship between 19 types of syndromes and 147 symptoms.The preliminary accuracy obtained through testing of syndrome differentiation reasoning can reach 79.17%,and the highest accuracy achieved by grouping statistics according to syndrome types can reach 100%.The syndrome differentiation results can be preliminarily explained based on the knowledge graph of depression intelligent assisted syndrome differentiation.This paper applies knowledge graph to the knowledge representation of intelligent assisted syndrome differentiation of TCM,and combines probabilistic reasoning methods for syndrome differentiation reasoning,which helps improve the efficiency of constructing syndrome differentiation models and the visualization degree of syndrome differentiation knowledge in the models.