通讯作者: 吕艳华, Lvyanhua01@163.com

DOI：10.12201/bmr.202604.00081

Theme Mining and Evolutionary Analysis of Medical Artificial Intelligence Research in China Over the Past Decade from the Perspectives of SIFRank and DTM

• 摘要：目的/意义 医学人工智能在促进医学信息化与智能化发展中具有关键作用，本文梳理我国近十年医学人工智能领域的研究文献，揭示其主要研究主题与发展趋势，为后续研究与实践提供数据支撑与方向参考。方法/过程 以中国知网、万方和维普数据库中2015年1月1日至2025年9月10日期间发表的7059篇医学人工智能相关文献为研究对象，提取题目、关键词及摘要作为分析语料，采用融合SIFRank与DTM模型的方法进行主题挖掘，结合主题热度与主题相似度指标，分析三个时间阶段的研究热点及其演化特征。结果/结论 共识别出10个医学人工智能领域的核心研究主题，并揭示了该领域的演化路径呈现出以肿瘤学为核心应用领域、传统医学数智化形成持续性发展脉络、研究趋于体系化发展三个演化特征。总体而言，医学人工智能研究当前已进入并将在长期内处于深度融合阶段，未来该领域的发展不仅要追求技术融入国内医学体系的深度与广度，更要直面国内临床实际中的难点与痛点，破解落地困境，形成持续创新且务实可行的发展路径。

  关键词： 医学人工智能；SIFRank-DTM融合模型；主题挖掘；热点识别；主题演化; ; ; ; 

   

  Abstract: Abstract Purpose/Significance Medical artificial intelligence drives advancements in medical informatization and intelligent development. This paper reviews research literature in the field of medical AI in China over the past decade, revealing its primary research themes and developmental trends to provide data support and directional guidance for subsequent research and practice. Method/Process This study analyzed 7,059 medical AI-related publications from January 1, 2015, to September 10, 2025, in the CNKI, Wanfang, and VIP databases. Titles, keywords, and abstracts were extracted as the analysis corpus. The theme mining process began with using SIFRank to extract key phrases from the corpus, and then the Dynamic Topic Model (DTM) was applied to identify and track themes over time. Theme popularity and similarity metrics were integrated to examine research hotspots and their evolutionary characteristics across three distinct time periods. Result/Conclusion Ten core research themes in medical artificial intelligence have been identified, revealing three key evolutionary characteristics: oncology as the primary application domain, the continuous digital transformation of traditional medicine, and the trend toward systematic research development. Overall, medical AI research has now entered and will remain in a phase of deep integration for the foreseeable future. Future development in this field must not only pursue the depth and breadth of technological integration into the domestic medical system, but alsodirectly address the challenges and pain points encountered in clinical practice. By overcoming implementation barriers, it will forge a path of continuous innovation that is both pragmatic and feasible.

  Key words: Medical Artificial Intelligence; SIFRank-DTM Hybrid Model; Topic Mining; Hotspot Identification; Topic Evolution

  提交时间：2026-04-10

  版权声明：作者本人独立拥有该论文的版权，预印本系统仅拥有论文的永久保存权利。任何人未经允许不得重复使用。
• html

• 图表

• 于诗睿, 李爱花, 林紫洛, 陈逸菲, 唐小利. 基于主题模型的科技文献主题演化分析方法优化研究综述. 2023. doi: 10.12201/bmr.202305.00016

  董怡, 冉晔, 余中光. 我国医疗人工智能风险研究现状及风险识别. 2024. doi: 10.12201/bmr.202411.00081

  张文丽, 沈丽宁. 负责任研究与创新视角下医学人工智能研究的伦理责任探究. 2026. doi: 10.12201/bmr.202605.00072

  李志远, 季惠斌, 李姗, 吴华章. 新医改以来公立医院补偿机制改革政策的主题识别与演化分析. 2025. doi: 10.12201/bmr.202508.00042

  王涟, 袁永旭, 陈俊冶, 殷彩明, 孙一凡. 基于LDA模型的我国中医药政策主题演化分析. 2024. doi: 10.12201/bmr.202404.00039

  解静, 刘久畅. 人工智能时代美国医学教育改革前沿案例分析及其启示. 2025. doi: 10.12201/bmr.202503.00054

  赵锐, 石秀园, 钟雪然, 任平, 田雪晴, 刘春平, 游茂. 基于卫生技术评估构建医学人工智能技术临床应用评估路径的探讨. 2022. doi: 10.12201/bmr.202203.00005

  黄晗琦, 周曼慈, 杨惠雯, 王睿智, 王若佳. 社会公众视角下医疗人工智能应用的伦理风险研究*. 2025. doi: 10.12201/bmr.202511.00062

  李帅祥, 施子文, 石慧敏. 基于CiteSpace的国内外人工智能在甲状腺疾病研究的可视化分析. 2026. doi: 10.12201/bmr.202601.00084

  弓孟春, 李雨杭, 马永慧, 弓凯, 刘超, 欧阳自豪, 戴辉. 生成式医学人工智能(GMAI)的临床伦理治理:三维协同路径与中国实践. 2025. doi: 10.12201/bmr.202511.00057

• 序号	提交日期	编号	操作
  1	2025-12-13	10.12201/bmr.202604.00081V1	下载

• 公开评论  匿名评论  仅发给作者 提交评论