基于Neo4j的中风病知识图谱构建思路与应用探析

Discussion on the construction and application of stroke knowledge graph based on Neo4j

目的:运用图数据库Neo4j Community 3.5.25初步构建中医医案中风病知识图谱,探索名老中医中风病知识图谱构建与应用的标准化流程。方法:以《中国百年百名中医临床家丛书》《中国现代名中医医案精粹》中治疗中风病的医案资料为基础,采用实体链指、关系抽取、知识表示及融合、知识存储及构建技术,借鉴SNOMED、UMLS和TCMLS的语义关系科学,人工整理构建素材并将构建素材以表格形式存储于WPS office。利用SQL Server 2019对数据进行预处理,在Neo4j Browser中运用Cypher语言以关联知识图谱的形式将“证-症”“证-法”“证-方-药”“症-方-药”“症-证-法”的关系予以展示;在图数据库Neo4j中构建中风病关系的语义网络支持系统。结果:共收集到中风病医案资料217份,初步构建中风病知识图谱,其中包含实体1853个和关联5781个。得到“证-症”“证-法”“证-方-药”“症-方-药”“症-证-法”中风病分类知识图谱。结论:本研究清晰地展示了“症-证-法-方-药”之间复杂的网络关系,展现了辨病与辨证相结合的思路并为在线诊治提供方法,还为发现中风病知识图谱与复杂网络的共性及探索运用它们的普遍方法奠定了基础。

Objective:To construct the stroke knowledge graph of TCM medical records by using the map database Neo4j community 3.5.25,and explore the standardized process of the construction and application of the famous Traditional Chinese Medicine stroke knowledge graph.On the basis of the medical case data of treating stroke in A series of 100 Chinese medicine clinicians in the past century and The quintessence of modern famous TCM doctor medical records in China,the construction materials were manually sorted out and stored in WPS office in the form of tables by using the technologies of entity chain index,relationship extraction,knowledge representation and fusion,knowledge storage and construction,etc..SQL Server 2019 was used to pre process the data,and Cypher language was used in Neo4j browser to display the relationship of“syndrome-symptom”“syndrome-method”“symptom-prescription-medicine”“syndrome-prescription-medicine”“symptom-syndrome-method”in the form of relational knowledge map.A semantic network support system for stroke relationship was constructed in the graph database Neo4j.Results:A total of 217 medical records of stroke were collected,and a knowledge graph of stroke was initially constructed,including 1853 entities and 5781 associations.The classification knowledge graph of“syndrome-symptom”“syndrome-method”“syndrome-prescription-medicine”“symptom-prescription-medicine”and“syndrome-symptom-method”were obtained.Conclusion:This study clearly shows the complex network relationship between“syndrome-symptom-method-prescription-medicine”,shows the idea of combining disease differentiation and syndrome differentiation,and provides methods for online diagnosis and treatment.It also lays a foundation for discovering the commonalities of stroke knowledge graph and complex networks and exploring common methods.