| 注册
首页|期刊导航|中华中医药学刊|ChatGPT在风湿科中医电子病历症状信息预处理中的应用

ChatGPT在风湿科中医电子病历症状信息预处理中的应用

喻金龙 张磊 许宁 法立峰 杨扩 韩真真 郭洪涛

中华中医药学刊2025,Vol.43Issue(3):24-29,后插43,7.
中华中医药学刊2025,Vol.43Issue(3):24-29,后插43,7.DOI:10.13193/j.issn.1673-7717.2025.03.005

ChatGPT在风湿科中医电子病历症状信息预处理中的应用

Application of ChatGPT in Symptom Information Preprocessing of Electronic Medical Record of Rheumatology

喻金龙 1张磊 2许宁 2法立峰 2杨扩 3韩真真 4郭洪涛4

作者信息

  • 1. 中国中医科学院中医药信息研究所,北京 100700
  • 2. 中国中医科学院中医药数据中心,北京 100700
  • 3. 北京交通大学计算机与信息技术学院,北京 100081
  • 4. 河南中医药大学第一附属医院,河南郑州 450000
  • 折叠

摘要

Abstract

Objective Information extraction is an important means of natural language processing.Based on the natural lan-guage processing ability of ChatGPT,this paper extracted the symptom information from traditional Chinese medicine electronic medical records in rheumatism department through ChatGPT.Methods Through the small sample learning based on ChatGPT large model,the information extraction and standardization of chief complaint and present medical history,specialist examination and tongue and pulse data in rheumatic electronic medical records were realized.At the same time,a large model ChatGLM was selected for the same small sample learning as ChatGPT,the same information extraction task was completed and the accuracy of the extraction results of the two large models was compared.Results Aiming at the problem of medical named entity recognition,the accuracy of ChatGPT after small sample learning reached 98.7%in chief complaint symptom extraction task,92%in special-ist examination and extraction task,and 98%in tongue and pulse image extraction and standardization task.The correct rates of ChatGLM on the same three tasks were 85.7%,91%and 98%respectively.From the extraction results,the two large models had different defects in the extraction task,but on the whole,ChatGPT performed better.Conclusion This paper realized the TCM symptom information preprocessing based on ChatGPT small sample learning,which was more convenient and efficient than the traditional model,and performed better than ChatGLM as a whole,which provided a new idea for TCM clinical text information ex-traction research.

关键词

ChatGPT/中医电子病历/症状/数据预处理/信息抽取

Key words

ChatGPT/TCM electronic medical records/symptoms/data preprocessing/information extraction

分类

医药卫生

引用本文复制引用

喻金龙,张磊,许宁,法立峰,杨扩,韩真真,郭洪涛..ChatGPT在风湿科中医电子病历症状信息预处理中的应用[J].中华中医药学刊,2025,43(3):24-29,后插43,7.

基金项目

国家重点研发计划项目(2017YFC1703504) (2017YFC1703504)

中央级公益性科研院所基本科研业务费专项资金资助项目(ZZ16-XRZ-108-SJ) (ZZ16-XRZ-108-SJ)

中国中医科学院科技创新工程项目(CI2021A05404) (CI2021A05404)

河南省中医药科学研究专项课题项目(2021JDZx2010) (2021JDZx2010)

中华中医药学刊

OA北大核心

1673-7717

访问量0
|
下载量0
段落导航相关论文