基于大语言模型的医疗质量控制应用系统构建  

Construction of a Medical Quality Control Application System Based on Large Language Models

作　　者：吴宏 胡军 陈尔真 董晨杰 李建华[3] 叶琪[3] WU Hong;HU Jun;CHEN Erzhen;DONG Chenjie;LI Jianhua;YE Qi(Shanghai Municipal Health Commission,Shanghai 200125,China;Shanghai Medical Quality Control Management Center,Shanghai 200040,China;East China University of Science and Technology,Shanghai 200237,China)

机构地区：[1]上海市卫生健康委员会,上海200125 [2]上海市医疗质量控制管理事务中心,上海200040 [3]华东理工大学,上海200237

出　　处：《医学信息学杂志》2025年第2期8-13,共6页Journal of Medical Informatics

基　　金：国家重点研发计划(项目编号:2023YFF1204904)。

摘　　要：目的/意义开发基于大语言模型的医疗质量控制系统,以提升医疗质量控制自动化水平与准确性。方法/过程获取高质量医学数据(包括通用医疗数据和质量控制相关数据)对PULSE模型微调,结合自动化指标计算与检索增强生成技术,高效执行复杂质量控制任务。结果/结论该系统在医疗质量控制的自动化计算与推理准确性方面性能显著提升,准确度达93.31%,优于基座大模型和常见推理方法,具有重要的应用价值。Purpose/Significance To develop a medical quality control system based on large language models(LLM),in order to enhance the automation and accuracy of quality control processes.Method/Process The system utilizes a diverse set of high-quality medical data,including general medical data and quality control-related data,to fine tune the PULSE model.Automated indicator calculation and retrieval-augmented generation techniques are employed to enable the system to efficiently perform complex quality control tasks.Result/Conclusion The system demonstrates significant improvements in automated computation and inference accuracy,achieving an accuracy of 93.31%.It outperforms baseline language models and common inference methods,offering significant implications for medical quality control.

关 键 词：大语言模型 指令微调 医疗质量控制 

分 类 号：R-058[医药卫生]