通讯作者: 余晓莹, bmxiaoyingy@mail.scut.edu.cn

DOI：10.12201/bmr.202609.00006

Research on Public Opinion of Medical Artificial Intelligence Based on BERTopic

• 摘要：摘要目的/意义 揭示公众对医疗人工智能的认知态度、核心关注问题及动态演化规律，为政策制定和治理优化提供决策支持，推动医疗AI普惠发展。方法/过程 采集抖音（23 862条）、小红书（5 085条）和微博（1 866条）医疗人工智能相关帖文评论，经预处理获得29 424条有效评论，采用BERTopic模型开展主题识别与演化分析，并结合BERT模型进行情感分析。结果/结论 共识别出AI诊疗能力边界认知、AI医疗职业冲击感知、AI医疗应用落地条件、AI医疗责任归属争议、AI医疗普惠价值期待和AI医疗未来影响预期六个核心议题；公众情感整体呈现“消极主导、两极分化”特征，不同平台间情感分布存在明显差异。本研究克服了传统方法在短文本分析中的局限，创新性地构建了从“理论预设”转向“公众自发”的舆情分析新视角，为识别公众真实关切、助推“健康中国”建设提供了新范式。

  关键词： 医疗人工智能；BERTopic；主题演化；平台差异；公众认知; ; ; ; 

   

  Abstract: Purpose/Significance This study reveals the public’s cognitive attitude, core concerns and dynamic evolution of medical artificial intelligence, provides decision support for policy formulation and governance optimization, and promotes the inclusive development of medical AI. Method/Process This study collected Douyin ( 23 862 ), Xiaohongshu ( 5 085 ) and Weibo ( 1 866 ) medical artificial intelligence related post comments, and obtained 29 424 valid comments after preprocessing. The BERTopic model was used to carry out topic recognition and evolution analysis, and the BERT model was used for sentiment analysis.Result/Conclusion Six core issues were identified, including boundary cognition of AI diagnosis and treatment ability, perception of AI medical occupational shock, landing conditions of AI medical application, dispute of AI medical responsibility attribution, expectation of AI medical inclusive value and expectation of AI medical future impact. The overall public sentiment is characterized by '' negative dominance and polarization '', and there are obvious differences in the distribution of emotions between different platforms. This study overcomes the limitations of traditional methods in short text analysis, and innovatively constructs a new perspective of public opinion analysis from “theoretical presupposition” to “public spontaneity”, which provides a new paradigm for identifying the real concerns of the public and promoting the construction of “Healthy China”.

  Key words: Medical Artificial Intelligence; BERTopic; Topic Evolution; Platform Differences; Public Perception

  提交时间：2026-09-04

  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

• 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

  中国科学技术信息研究所, 北京市海淀区复兴路号. 基于生成式人工智能的医学视频创作路径探究. 2025. doi: 10.12201/bmr.202508.00005

  张世红, 史森, 白玲, 杨小冉, 韩冬, 胡红濮. 人工智能在公共卫生应急管理的应用场景分析及挑战. 2025. doi: 10.12201/bmr.202511.00063

  吕艳华, 王萱, 崔云龙, 朱鑫鑫, 白慧娟. 融合SIFRank和DTM视角下近十年我国医学人工智能研究主题挖掘及其演化分析. 2026. doi: 10.12201/bmr.202604.00081

  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

  马傲, 葛小玲. 人工智能大模型在医疗健康领域应用的研究. 2024. doi: 10.12201/bmr.202408.00039

  李明. 医生对医疗人工智能使用意愿性以及影响因素分析. 2023. doi: 10.12201/bmr.202312.00020

  孟冬晴, 吕文娟. 生成式人工智能赋能医学图书馆知识服务的路径研究. 2025. doi: 10.12201/bmr.202510.00039

  郑琰莉, 韩福海, 李舒玉, 苏文星. 人工智能大模型在医疗领域的应用现状与前景展望. 2023. doi: 10.12201/bmr.202312.00027

  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

• 序号	提交日期	编号	操作
  1	2026-05-13	10.12201/bmr.202609.00006V1	下载

• 公开评论  匿名评论  仅发给作者 提交评论