{"title":"中国的电子烟管制政策有效吗?使用人工智能集成混合方法对公共话语进行长达十年的分析。","authors":"Zhangyan Li, Xinrui Wang, Xingye Yao, Yu Chen","doi":"10.18332/tid/208810","DOIUrl":null,"url":null,"abstract":"<p><strong>Introduction: </strong>China, the world's largest tobacco market, has raised concerns due to e-cigarettes' health risks and rising youth usage. Despite a decade of regulatory policies, their effectiveness remains uncertain. This study examines trends in e-cigarette discourse on Weibo (2016-2025), analyzing discussion volume shifts and the impact of various topics on public engagement.</p><p><strong>Methods: </strong>This study employs a hybrid computational approach integrating topic modeling, LLM-assisted annotation, and quantitative analysis to examine the evolution of e-cigarette discussions on Weibo (2015-2025) and topic dissemination effects (n=129769). LDA modeling identify 10 topics, followed by DeepSeek-V3-assisted classification. Linear regression in SPSS analyzed relationships between topic categories and social media engagement metrics (reposts/comments/likes) at 95% confidence intervals.</p><p><strong>Results: </strong>Findings reveal 2020 as a key year of change: pro-vaping posts declined while anti-vaping content increased. Despite reduced volume, pro-vaping material maintained significant digital influence. Pre-policy, marketing content (p<0.01), health effects (p<0.01) and regulation (p<0.01) drove engagement. Post-policy, marketing lost engagement impact, while 'user experience' posts gained traction, significantly correlating with all interactions (all p<0.05). This indicates regulations were less effective against user-generated content, with pro-vaping messaging shifting towards peer-driven channels. Crucially, influencers consistently triggered strong engagement throughout the period (p<0.01) despite lower post volume, remaining key discourse drivers.</p><p><strong>Conclusions: </strong>Although China is strengthening its control over e-cigarettes, the results of our study indicate that this control remains limited. We advocate for more robust regulation of social media content, particularly concerning the management of celebrities and influencers, as well as the sharing of e-cigarette use experiences. However, the current regulatory framework enforced by the State Tobacco Monopoly Administration has proven inadequate for widespread and effective governance. We suggest that regulatory authority be shared with public health agencies in order to better integrate e-cigarette regulation with broader public health objectives.</p>","PeriodicalId":23202,"journal":{"name":"Tobacco Induced Diseases","volume":"23 ","pages":""},"PeriodicalIF":1.9000,"publicationDate":"2025-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12465113/pdf/","citationCount":"0","resultStr":"{\"title\":\"Do China's e-cigarette control policies work? A decade-long analysis of public discourse using an AI-integrated mixed-methods approach.\",\"authors\":\"Zhangyan Li, Xinrui Wang, Xingye Yao, Yu Chen\",\"doi\":\"10.18332/tid/208810\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Introduction: </strong>China, the world's largest tobacco market, has raised concerns due to e-cigarettes' health risks and rising youth usage. Despite a decade of regulatory policies, their effectiveness remains uncertain. This study examines trends in e-cigarette discourse on Weibo (2016-2025), analyzing discussion volume shifts and the impact of various topics on public engagement.</p><p><strong>Methods: </strong>This study employs a hybrid computational approach integrating topic modeling, LLM-assisted annotation, and quantitative analysis to examine the evolution of e-cigarette discussions on Weibo (2015-2025) and topic dissemination effects (n=129769). LDA modeling identify 10 topics, followed by DeepSeek-V3-assisted classification. Linear regression in SPSS analyzed relationships between topic categories and social media engagement metrics (reposts/comments/likes) at 95% confidence intervals.</p><p><strong>Results: </strong>Findings reveal 2020 as a key year of change: pro-vaping posts declined while anti-vaping content increased. Despite reduced volume, pro-vaping material maintained significant digital influence. Pre-policy, marketing content (p<0.01), health effects (p<0.01) and regulation (p<0.01) drove engagement. Post-policy, marketing lost engagement impact, while 'user experience' posts gained traction, significantly correlating with all interactions (all p<0.05). This indicates regulations were less effective against user-generated content, with pro-vaping messaging shifting towards peer-driven channels. Crucially, influencers consistently triggered strong engagement throughout the period (p<0.01) despite lower post volume, remaining key discourse drivers.</p><p><strong>Conclusions: </strong>Although China is strengthening its control over e-cigarettes, the results of our study indicate that this control remains limited. We advocate for more robust regulation of social media content, particularly concerning the management of celebrities and influencers, as well as the sharing of e-cigarette use experiences. However, the current regulatory framework enforced by the State Tobacco Monopoly Administration has proven inadequate for widespread and effective governance. We suggest that regulatory authority be shared with public health agencies in order to better integrate e-cigarette regulation with broader public health objectives.</p>\",\"PeriodicalId\":23202,\"journal\":{\"name\":\"Tobacco Induced Diseases\",\"volume\":\"23 \",\"pages\":\"\"},\"PeriodicalIF\":1.9000,\"publicationDate\":\"2025-09-26\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12465113/pdf/\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Tobacco Induced Diseases\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.18332/tid/208810\",\"RegionNum\":4,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"2025/1/1 0:00:00\",\"PubModel\":\"eCollection\",\"JCR\":\"Q2\",\"JCRName\":\"PUBLIC, ENVIRONMENTAL & OCCUPATIONAL HEALTH\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Tobacco Induced Diseases","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.18332/tid/208810","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/1/1 0:00:00","PubModel":"eCollection","JCR":"Q2","JCRName":"PUBLIC, ENVIRONMENTAL & OCCUPATIONAL HEALTH","Score":null,"Total":0}
Do China's e-cigarette control policies work? A decade-long analysis of public discourse using an AI-integrated mixed-methods approach.
Introduction: China, the world's largest tobacco market, has raised concerns due to e-cigarettes' health risks and rising youth usage. Despite a decade of regulatory policies, their effectiveness remains uncertain. This study examines trends in e-cigarette discourse on Weibo (2016-2025), analyzing discussion volume shifts and the impact of various topics on public engagement.
Methods: This study employs a hybrid computational approach integrating topic modeling, LLM-assisted annotation, and quantitative analysis to examine the evolution of e-cigarette discussions on Weibo (2015-2025) and topic dissemination effects (n=129769). LDA modeling identify 10 topics, followed by DeepSeek-V3-assisted classification. Linear regression in SPSS analyzed relationships between topic categories and social media engagement metrics (reposts/comments/likes) at 95% confidence intervals.
Results: Findings reveal 2020 as a key year of change: pro-vaping posts declined while anti-vaping content increased. Despite reduced volume, pro-vaping material maintained significant digital influence. Pre-policy, marketing content (p<0.01), health effects (p<0.01) and regulation (p<0.01) drove engagement. Post-policy, marketing lost engagement impact, while 'user experience' posts gained traction, significantly correlating with all interactions (all p<0.05). This indicates regulations were less effective against user-generated content, with pro-vaping messaging shifting towards peer-driven channels. Crucially, influencers consistently triggered strong engagement throughout the period (p<0.01) despite lower post volume, remaining key discourse drivers.
Conclusions: Although China is strengthening its control over e-cigarettes, the results of our study indicate that this control remains limited. We advocate for more robust regulation of social media content, particularly concerning the management of celebrities and influencers, as well as the sharing of e-cigarette use experiences. However, the current regulatory framework enforced by the State Tobacco Monopoly Administration has proven inadequate for widespread and effective governance. We suggest that regulatory authority be shared with public health agencies in order to better integrate e-cigarette regulation with broader public health objectives.
期刊介绍:
Tobacco Induced Diseases encompasses all aspects of research related to the prevention and control of tobacco use at a global level. Preventing diseases attributable to tobacco is only one aspect of the journal, whose overall scope is to provide a forum for the publication of research articles that can contribute to reducing the burden of tobacco induced diseases globally. To address this epidemic we believe that there must be an avenue for the publication of research/policy activities on tobacco control initiatives that may be very important at a regional and national level. This approach provides a very important "hands on" service to the tobacco control community at a global scale - as common problems have common solutions. Hence, we see ourselves as "connectors" within this global community.
The journal hence encourages the submission of articles from all medical, biological and psychosocial disciplines, ranging from medical and dental clinicians, through health professionals to basic biomedical and clinical scientists.