摘要
目的 构建一个融合大语言模型与知识图谱的抑郁症中西医结合智能问答系统,通过精准的问题处理、高效的知识检索及专业的回答生成,为抑郁症患者及医疗人员提供高效、准确的诊疗支持,推动抑郁症诊疗的智能化发展。方法 基于《抑郁症中西医结合诊疗指南》,利用py2neo库与Neo4j图数据库,构建以“病-证-症-治-法-方-药”为核心的抑郁症知识图谱,并通过模板匹配技术,将自然语言问题转换为结构化查询,结合Cypher语句和大语言模型(ChatGLM4)生成专业回答。结果 所构建的抑郁症知识图谱包含了303个实体节点以及436条实体关系,其模式层包含19种实体类型、14种关系类型,成功实现了数据的可视化,能够辅助临床医生直观查看和检索诊疗数据。该系统能够准确处理抑郁症相关问题,提供高效和精准的回答。结论 该系统在抑郁症诊疗中展现出良好的应用潜力,可辅助临床医生进行疾病诊疗和临床决策,有助于诊疗指南的知识共享、推广以及诊疗过程的规范化和标准化,该知识图谱和智能问答系统的构建思路也可为其他疾病的智能诊疗提供参考和借鉴。
Objective To construct an intelligent question-answering system that integrates large language models and knowledge graphs for depression treatment combining traditional Chinese medicine and Western medicine.The question-answering system aims to provide efficient and accurate diagnostic and therapeutic support for both patients and healthcare professionals through precise question processing,efficient knowledge retrieval,and professional answer generation,thereby promoting the intelligent development of depression diagnosis and treatment.Methods Based on the Guidelines for Integrating Traditional Chinese and Western Medicine in the Diagnosis and Treatment of Depression,a depression knowledge graph centered on"disease-syndrome-symptom-treatment-method-prescription-medicine"was constructed using the py2neo library and the Neo4j graph database.Natural language questions were converted into structured queries through template matching technology,and professional answers were generated by combining Cypher queries and a large language model(ChatGLM4).Results The constructed depression knowledge graph included 303 entity nodes and 436 entity relationships,and its schema layer contained 19 types of entities and 14 types of relationships.Therefore,the system successfully visualized the data,enabling clinical doctors to visually view and retrieve diagnostic and treatment data.It accurately processed depression-related questions,providing efficient and precise answers.Conclusions The system demonstrates significant application potential in the diagnosis and treatment of depression.It can assist clinical doctors in disease diagnosis,treatment,and clinical decision-making,facilitating knowledge sharing and dissemination,as well as standardization of diagnostic and treatment process.The approach used to construct the knowledge graph and intelligent question-answering system can serve as a reference for the intelligent diagnosis and treatment of other diseases.
作者
谭平
刘惠娜
韦昌法
TAN Ping;LIU Huina;WEI Changfa(School of Informatics and Engineering,Hunan University of Chinese Medicine,Changsha,Hunan 410208,China;Medical School,Hunan University of Chinese Medicine,Changsha,Hunan 410208,China;Hunan Smart Traditional Chinese Medicine Engineering Technology Research Center,Changsha,Hunan 410208,China)
出处
《上海中医药杂志》
2025年第7期1-10,共10页
Shanghai Journal of Traditional Chinese Medicine
基金
湖南省自然科学基金项目(2025JJ90089)
湖南省研究生科研创新项目(CX20240735)
湖南省研究生科研创新立项不资助项目(LXBZZ2024175)
湖南中医药大学研究生创新课题(2024CX084)。
关键词
人工智能
大语言模型
知识图谱
抑郁症
智能问答
诊疗指南
artificial intelligence
large language model
knowledge graph
depression
intelligent question-answering
diagnosis and treatment guidelines
作者简介
谭平,男,硕士研究生,主要从事医学自然语言处理研究工作;通信作者:刘惠娜,讲师,E-mail:huinaliu@hnucm.edu.cn。;通信作者:韦昌法,教授,硕士研究生导师,E-mail:weichangfa@hnucm.edu.cn。
