Department of Psychiatry, Washington University School of Medicine, 660 S Euclid Ave, St. Louis, MO, USA.
Department of Computer Science and Engineering, University of South Carolina, Columbia, SC, USA.
Int J Med Inform. 2021 Nov;155:104574. doi: 10.1016/j.ijmedinf.2021.104574. Epub 2021 Sep 14.
Vaping product use (i.e., e-cigarettes) has been rising since 2000 in the United States. Negative health outcomes associated with vaping products have created public uncertainty and debates on social media platforms. This study explores the feasibility of using social media as a surveillance tool to identify relevant posts and at-risk vaping users.
Using an interdisciplinary method that leverages natural language processing and manual content analysis, we extracted and analyzed 794,620 vaping-related tweets on Twitter. After observing significant increases in vaping-related tweets in July, August, and September 2019, additional human coding was completed on a subset of these tweets to better understand primary themes of vaping-related discussions on Twitter during this time frame.
We found significant increases in tweets related to negative health outcomes such as acute lung injury and respiratory issues during the outbreak of e-cigarette/vaping associated lung injury (EVALI) in the fall of 2019. Positive sentiment toward vaping remained high, even across the peak of this outbreak in July, August, and September. Tweets mentioning the public perceptions of youth risk were concerning, as were increases in marketing and marijuana-related tweets during this time.
The preliminary results of this study suggest the feasibility of using Twitter as a means of surveillance for public health crises, and themes found in this research could aid in specifying those groups or populations at risk on Twitter. As such, we plan to build automatic detection algorithms to identify these unique vaping users to connect them with a digital intervention in the future.
自 2000 年以来，美国的蒸气产品使用（即电子烟）一直在增加。与蒸气产品相关的负面健康后果在社交媒体平台上引发了公众的不确定性和争论。本研究探讨了利用社交媒体作为监测工具识别相关帖子和有风险蒸气使用者的可行性。
我们使用一种跨学科的方法，利用自然语言处理和手动内容分析，从 Twitter 上提取和分析了 794620 条与蒸气相关的推文。在 2019 年 7 月、8 月和 9 月观察到与蒸气相关的推文显著增加后，我们对这些推文的一个子集进行了额外的人工编码，以更好地了解这段时间内 Twitter 上与蒸气相关讨论的主要主题。
我们发现，在 2019 年秋季电子烟/蒸气相关肺损伤（EVALI）爆发期间，与负面健康后果（如急性肺损伤和呼吸问题）相关的推文显著增加。即使在 7 月、8 月和 9 月这一疫情高峰期，对蒸气的积极情绪仍然很高。提到公众对年轻人风险的看法令人担忧，在这段时间里，市场营销和大麻相关的推文也有所增加。
本研究的初步结果表明，利用 Twitter 作为公共卫生危机监测手段是可行的，并且本研究中发现的主题可以帮助确定 Twitter 上的那些处于危险中的群体或人群。因此，我们计划建立自动检测算法，以识别这些独特的蒸气使用者，以便将来与他们建立数字干预。
