基于本地大模型的骨质疏松专病数据库构建流程优化研究  

Study on Optimization of Osteoporosis Specialized Disease Database Construction Process Based on Local Large Models

作　　者：王耀国 唐诗诗 刘泓泽 安雨婷 周毅[2] WANG Yaoguo;TANG Shishi;LIU Hongze;AN Yuting;ZHOU Yi(School of Public Health,Xinjiang Medical University,Urumqi 830017,China;Zhongshan School of Medicine,Sun Yat-sen University,Guangzhou 510080,China)

机构地区：[1]新疆医科大学公共卫生学院,乌鲁木齐830017 [2]中山大学中山医学院,广州510080

出　　处：《医学信息学杂志》2025年第1期75-79,共5页Journal of Medical Informatics

基　　金：国家重点研发计划(项目编号:2022YFC3601600);广东省自然科学基金项目(项目编号:2024A1515011989);新疆维吾尔自治区人民医院院内课题(项目编号:20230117)。

摘　　要：目的/意义利用本地大模型和知识库优化骨质疏松专病数据库构建流程,降低医疗数据泄漏风险。方法/过程收集相关临床研究文献和业务数据信息,构建本地知识库;基于本地大模型应用架构,使用本地知识库限定本地大模型的生成内容;基于本地大模型应用架构图形问答界面和编程接口,通过自然语言问询和编程引导本地大模型生成骨质疏松专病数据库字段集。结果/结论本地大模型结合本地知识库可以总结骨质疏松既往研究中涉及的字段,有效提高骨质疏松专病数据库构建效率和质量,但处理结果需要人工核查。Purpose/Significance To utilize local large models combined with local knowledge base to optimize the osteoporosis disease database construction process,and to reduce the risk of medical data leakage.Method/Process Firstly,relevant clinical research literature and business data information are collected and a local knowledge base is built.Then,based on the local large model application architecture,the local knowledge base is used to limit the generated content of the local large model.Finally,based on the graphical question-and-answer interface and programming interface of the local large model application architecture,natural language queries and programming methods are used to guide the local large model to generate osteoporosis disease database field sets and data production.Result/Conclusion The local large models combined with the local knowledge base can summarize the fields involved in previous osteoporosis researches and effectively improve the efficiency and quality of osteoporosis disease database construction,but the processing results need to be manually verified.

关 键 词：专病数据库 大模型 本地化部署 自然语言理解 非结构化数据处理 

分 类 号：R-058[医药卫生]