文献速递中药与中成药合理使用

基于文本挖掘的微信公众号药学科普传播策略研究

Research on strategies for enhancing drug knowledge dissemination on Chinese social media WeChat public accounts based on text mining technology.
PubMed 文献 · Front Pharmacol · 2025-08-26

该研究以知名药学科普微信公众号「PSM药盾公益」为例,结合TF-IDF算法与VOSviewer可视化分析构建药学科普文章热点主题分析模型,并采用LDA与BERTopic模型实现主题建模。研究选取2015至2023年间阅读量前20%的科普文章作为文本挖掘数据库,聚类结果显示公众关注药品剂量、药品副作用、儿童感染、中医药及中成药疗效、不同给药途径使用方法五类主题,且关键词时间序列图显示公众兴趣从药品副作用转向实际用药问题。

要点

  • 公众对药学科普的关注集中于药品剂量、副作用、儿童感染、中医药疗效及给药途径五类主题。
  • TF-IDF结合VOSviewer及LDA、BERTopic文本挖掘可有效识别行业热点主题。
  • 公众兴趣随时间从药品副作用转向实际用药问题,科普选题应动态调整。

核心结论

  • 药学专业人员结合TF-IDF算法与VOSviewer可视化分析及LDA、BERTopic文本挖掘,可更有效发现行业热点主题,提升科普文章可读性与公众号影响力,进而可能改善用药依从性并提高公众用药意识。

研究设计与样本

  • 以微信公众号「PSM药盾公益」为研究平台,选取2015至2023年间阅读量前20%的科普文章作为文本挖掘数据库,结合TF-IDF算法与VOSviewer可视化分析构建热点主题分析模型,并采用LDA与BERTopic模型实现主题建模。

主要发现

  • 聚类结果显示公众对药品剂量、药品副作用、儿童感染、中医药及中成药疗效、不同给药途径使用方法五类药学科普主题感兴趣。
  • 关键词时间序列图显示公众对主题的兴趣从药品副作用转向实际用药使用问题。

临床怎么用

  • 药学专业人员可结合TF-IDF算法与VOSviewer可视化分析及LDA、BERTopic文本挖掘发现行业热点主题,提升科普文章可读性与公众号影响力。
  • 科普选题可参考公众关注的五类主题(药品剂量、副作用、儿童感染、中医药及中成药疗效、给药途径使用方法)及兴趣从副作用向实际用药问题转移的趋势。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

OBJECTIVE: Health science popularization is an important means to improve public health literacy, promote healthy lifestyles, prevent diseases and respond to health crises, which is of great significance for improving the overall health of the people. Strengthening the medication education of patients is also one of the key factors to improve patients' medication adherence. In order to strengthen the dissemination of pharmaceutical popular science articles and give full play to the value of pharmaceutical popular science, this study takes WeChat public account as the research platform to explore effective strategies to improve pageviews of science popularization. It provides references for science popularization workers, so that science popularization can play a better role in improving the public's knowledge of medication safety. METHODS: Taking the well-known pharmaceutical science popularization WeChat account "PSM Medicine Shield Public Welfare" as an example, we combined the Term Frequency-Inverse Document Frequency (TF-IDF) algorithm and VOSviewer visualization analysis technology to construct a hot topic analysis model for pharmaceutical science popularization articles, and analyzed the common rules and characteristics of successful hot articles. Latent Dirichlet Allocation (LDA) and The Bidirectional Encoder Representations from Transformers Topic (BERTopic) model were used to realize the construction of the topic model. RESULTS: The model selected the top 20% of popularization articles with the greatest reading volume between 2015 and 2023 as the database for text mining. The clustering results indicated that the public was interested in these five types of pharmaceutical science popularization themes: drug dosage, drug side effects, children's infections, the efficacy of traditional Chinese medicine and Chinese patent medicines, and the usage methods of different drug administration routes. The public's interest in topics changed from drug side effects to practical drug usage issues, as seen by the keyword time series graph. CONCLUSION: Pharmaceutical professionals may more effectively discover hot themes in the industry by combining the TF-IDF algorithm with VOSviewer visualization analysis and LDA and BERTopic in the text mining. This improves the readability of popularization articles and the impact of WeChat accounts, which may improve medication adherence and raise public awareness of medication usage.

来源:Front Pharmacol,查看原文。版权归原作者所有。

引用

  1. Yu X, Chen X, Yan X, et al. Research on strategies for enhancing drug knowledge dissemination on Chinese social media WeChat public accounts based on text mining technology. Front Pharmacol. 2025. PMID: 40978476. doi:10.3389/fphar.2025.1569863

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。