Study on Knowledge Map of Prescribing Experience Inheritance of Veteran TCM Physicians Based on Deep Learning
Objective To construct a knowledge map for the experience inheritance of veteran TCM physicians;To provide theoretical basis for the inheritance and innovation of academic experience of famous doctors.Methods Professor Zhang Yongkang's partial medical record data from the outpatient department of TCM of Shanxi Provincial People's Hospital from September 16,2019 to March 22,2021 were sorted out,and a dataset was constructed through manual annotation,and then were randomly divided into a training set(n=804)and a test set(n=345)at 7:3.TPLinker model was used for entity relationship joint extraction.Precision,Recall,and F1 value were set as the evaluation indicators of the model.Py2neo and Neo4j were used to construct a knowledge map of TCM medical records,and Cypher query sentences were used to realize the visualization process of knowledge map.Results The Precision of TPLinker model in the TCM medical record dataset was 0.886,the Recall was 0.977,and the F1 value was 0.929.The TCM medical record knowledge map contained 1 025 entities,and 4 428 relationships was constructed through the Py2neo database and stored in the Neo4j diagram database.Conclusion Knowledge map for the experience inheritance of veteran TCM physicians can fully tap into Professor Zhang's personalized thinking mode of syndrome differentiation,providing a new way to inherit and promote the academic ideas and clinical experience of veteran TCM physicians.
deep learningartificial intelligenceknowledge mapveteran TCM physiciansTPLinkerNeo4j
万方数据
维普
