中国全科医学 ›› 2025, Vol. 28 ›› Issue (25): 3200-3208.DOI: 10.12114/j.issn.1007-9572.2024.0377

• 数智医疗与信息化研究 • 上一篇    下一篇

基于CiteSpace的国内外医疗大语言模型研究热点演进及趋势分析

牛奔1, 朱晓倩2, 杨辰3, 梁万年4,*(), 刘珏5,*()   

  1. 1.518060 广东省深圳市,深圳大学医院管理研究院
    2.518060 广东省深圳市,深圳大学大湾区国际创新学院
    3.518060 广东省深圳市,深圳大学管理学院
    4.100084 北京市,清华大学万科公共卫生与健康学院
    5.100091 北京市,北京大学公共卫生学院
  • 收稿日期:2024-08-15 修回日期:2024-09-10 出版日期:2025-09-05 发布日期:2025-07-24
  • 通讯作者: 梁万年, 刘珏

  • 作者贡献:

    牛奔负责确定论文选题、文献筛选核对、论文撰写及修改、经费支持;朱晓倩、杨辰负责文献检索、筛选和去重,以及文献数据分析;刘珏负责图表制作、论文修改及质控、经费支持;梁万年负责论文选题、质控、指导及修改,并对文章整体负责。

  • 基金资助:
    国家自然科学基金重点项目(72334004); 国家自然科学基金优青项目(72122001); 广东省普通高校重点领域专项(2022ZDZX2054)

Evolution and Trends of Domestic and International Research Hotspots in the Field of Large Language Models in Medicine Based on CiteSpace

NIU Ben1, ZHU Xiaoqian2, YANG Chen3, LIANG Wannian4,*(), LIU Jue5,*()   

  1. 1. Hospital Management Institute, Shenzhen University, Shenzhen 518060, China
    2. Greater Bay Area International Institute for Innovation, Shenzhen University, Shenzhen 518060, China
    3. College of Management, Shenzhen University, Shenzhen 518060, China
    4. Vanke School of Public Health and Health, Tsinghua University, Beijing 100084, China
    5. School of Public Health, Peking University, Beijing 100091, China
  • Received:2024-08-15 Revised:2024-09-10 Published:2025-09-05 Online:2025-07-24
  • Contact: LIANG Wannian, LIU Jue

摘要: 背景 由于其强大的语言处理能力和广泛的应用潜力,以ChatGPT为代表的大语言模型引领了医疗领域自然语言处理的新趋势。 目的 本研究通过文献计量分析揭示2017年以来医疗大语言模型的研究热点、主题分布及未来发展方向。 方法 通过Web of Science、中国知网、万方数据知识服务平台和维普网数据库,系统检索和筛选2017年1月—2024年6月关于医疗大语言模型的文献。利用CiteSpace软件提取文献中的主题关键词等信息,分析并对比国内外研究的演进、热点和趋势。 结果 共纳入1 071篇相关文献,结果显示国外研究集中于人工智能、大语言模型、深度学习、知识图谱等技术在医学中的应用,而国内研究则相对较少,侧重于中文医学问答系统构建和医疗数据非结构化问题处理。 结论 深化医疗数据挖掘,拓展多场景应用,并借鉴国际大语言模型的微调和应用评估经验,促进我国医疗大语言模型技术的发展和医学领域应用。

关键词: 卫生保健提供, 医疗健康, 大语言模型, 文献计量分析, CiteSpace, 人工智能

Abstract:

Background

With advanced language processing abilities and broad potential application scope, large language models (LLMs) such as ChatGPT, are driving a new wave of natural language processing in the medical field.

Objective

This study aims to identify research hotspots, topic distribution, and future trends of medical LLMs using bibliometric analysis.

Methods

A systematic search was conducted across the Web of Science, CNKI, Wanfang Data, and VIP databases for literature on medical LLMs published between January 2017 and June 2024. CiteSpace software was used to extract thematic keywords and other information from the literature, analyze and compare the evolution, hotspots, and trends of domestic and international research.

Results

A total of 1 071 relevant papers were included, revealing that international research mainly focuses on applying artificial intelligence, LLMs, deep learning, and knowledge graphs in medicine. In contrast, domestic research is more limited, focusing on developing Chinese medical question-answering systems and solving unstructured medical data problems.

Conclusion

It is recommended to enhance medical data mining, broaden its application in various scenarios, and leverage international experiences in fine-tuning and evaluating LLMs to advance medical LLM development in China.

Key words: Delivery of health care, Healthcare, Large language models, Bibliometric analysis, CiteSpace, Artificial intelligence

中图分类号: