Identifying e-cigarette content on TikTok: Using a BERTopic Modeling approach.

INTRODUCTION: The use of hashtags is a common way to promote e-cigarette content on social media. Analysis of hashtags may provide insight into e-cigarette promotion on social media. However, the examination of text data is complicated by the voluminous amount of social media data. This study used machine learning approaches (i.e., Bidirectional Encoder Representations from Transformers [BERT] topic modeling) to identify e-cigarette content on TikTok. METHODS: We used 13 unique hashtags related to e-cigarettes (e.g., #vape) for data collection. The final analytic sample included 12,573 TikTok posts. To identify the best fitting number of topic clusters, we used both quantitative (i.e., coherence test) and qualitative approaches (i.e., researchers checked the relevance of text from each topic). We, then, grouped and characterized clustered text to each theme. RESULTS: We evaluated that N=18 was the ideal number of topic clusters. The 9 overarching themes were identified: Social media and TikTok-related features (N=4; "duet", "viral"), Vape shops and brands (N=3; "store"), Vape tricks (N=3; "ripsaw"), Modified use of e-cigarettes (N=1; "coil", "wire"), Vaping and girls (N=1; "girl"), Vape flavors (N=1; "flavors"), Vape and cigarettes (N=1; "smoke"), Vape identities and communities (N=1; "community"), and Non-English language (N=3; Romanian and Spanish). CONCLUSIONS: This study used a machine learning method, BERTopic modeling, to successfully identify relevant themes on TikTok. This method can inform future social media research examining other tobacco products, and tobacco regulatory policies such as monitoring of e-cigarette marketing on social media. IMPLICATIONS: This study can inform future social media research examining other tobacco products, and tobacco regulatory policies such as monitoring of e-cigarette marketing on social media.

Using Computer Vision to Detect E-cigarette Content in TikTok Videos.

INTRODUCTION: Previous research has identified abundant e-cigarette content on social media using primarily text-based approaches. However, frequently used social media platforms among youth, such as TikTok, contain primarily visual content, requiring the ability to detect e-cigarette-related content across large sets of videos and images. This study aims to use a computer vision technique to detect e-cigarette-related objects in TikTok videos. AIMS AND METHODS: We searched 13 hashtags related to vaping on TikTok (eg, #vape) in November 2022 and obtained 826 still images extracted from a random selection of 254 posts. We annotated images for the presence of vaping devices, hands, and/or vapor clouds. We developed a YOLOv7-based computer vision model to detect these objects using 85% of extracted images (N = 705) for training and 15% (N = 121) for testing. RESULTS: Our model's recall value was 0.77 for all three classes: vape devices, hands, and vapor. Our model correctly classified vape devices 92.9% of the time, with an average F1 score of 0.81. CONCLUSIONS: The findings highlight the importance of having accurate and efficient methods to identify e-cigarette content on popular video-based social media platforms like TikTok. Our findings indicate that automated computer vision methods can successfully detect a range of e-cigarette-related content, including devices and vapor clouds, across images from TikTok posts. These approaches can be used to guide research and regulatory efforts. IMPLICATIONS: Object detection, a computer vision machine learning model, can accurately and efficiently identify e-cigarette content on a primarily visual-based social media platform by identifying the presence of vaping devices and evidence of e-cigarette use (eg, hands and vapor clouds). The methods used in this study can inform computational surveillance systems for detecting e-cigarette content on video- and image-based social media platforms to inform and enforce regulations of e-cigarette content on social media.

The Normalization of Vaping on TikTok Using Computer Vision, Natural Language Processing, and Qualitative Thematic Analysis: Mixed Methods Study.

BACKGROUND: Social media posts that portray vaping in positive social contexts shape people's perceptions and serve to normalize vaping. Despite restrictions on depicting or promoting controlled substances, vape-related content is easily accessible on TikTok. There is a need to understand strategies used in promoting vaping on TikTok, especially among susceptible youth audiences. OBJECTIVE: This study seeks to comprehensively describe direct (ie, explicit promotional efforts) and indirect (ie, subtler strategies) themes promoting vaping on TikTok using a mixture of computational and qualitative thematic analyses of social media posts. In addition, we aim to describe how these themes might play a role in normalizing vaping behavior on TikTok for youth audiences, thereby informing public health communication and regulatory policies regarding vaping endorsements on TikTok. METHODS: We collected 14,002 unique TikTok posts using 50 vape-related hashtags (eg, #vapetok and #boxmod). Using the k-means unsupervised machine learning algorithm, we identified clusters and then categorized posts qualitatively based on themes. Next, we organized all videos from the posts thematically and extracted the visual features of each theme using 3 machine learning-based model architectures: residual network (ResNet) with 50 layers (ResNet50), Visual Geometry Group model with 16 layers, and vision transformer. We chose the best-performing model, ResNet50, to thoroughly analyze the image clustering output. To assess clustering accuracy, we examined 4.01% (441/10,990) of the samples from each video cluster. Finally, we randomly selected 50 videos (5% of the total videos) from each theme, which were qualitatively coded and compared with the machine-derived classification for validation. RESULTS: We successfully identified 5 major themes from the TikTok posts. Vape product marketing (1160/10,990, 8.28%) reflected direct marketing, while the other 4 themes reflected indirect marketing: TikTok influencer (3775/14,002, 26.96%), general vape (2741/14,002, 19.58%), vape brands (2042/14,002, 14.58%), and vaping cessation (1272/14,002, 9.08%). The ResNet50 model successfully classified clusters based on image features, achieving an average F1-score of 0.97, the highest among the 3 models. Qualitative content analyses indicated that vaping was depicted as a normal, routine part of daily life, with TikTok influencers subtly incorporating vaping into popular culture (eg, gaming, skateboarding, and tattooing) and social practices (eg, shopping sprees, driving, and grocery shopping). CONCLUSIONS: The results from both computational and qualitative analyses of text and visual data reveal that vaping is normalized on TikTok. Our identified themes underscore how everyday conversations, promotional content, and the influence of popular figures collectively contribute to depicting vaping as a normal and accepted aspect of daily life on TikTok. Our study provides valuable insights for regulatory policies and public health initiatives aimed at tackling the normalization of vaping on social media platforms.

Content Analysis of YouTube Videos Related to E-Cigarettes and COVID-19.

BACKGROUND: E-cigarettes are frequently promoted on social media and portrayed in ways that are attractive to youth. While the COVID-19 pandemic significantly affected people's lives, less is known about how the pandemic influenced e-cigarette-related marketing and information on social media. This study examined how e-cigarettes were portrayed on youtube, one of the most popular social media platforms during the COVID-19 pandemic. METHODS: We searched for combinations of search terms related to e-cigarettes (e.g., "electronic cigarette" and "vape") and COVID-19 (e.g., "corona" and "COVID") in July of 2021. To be included in analyses, videos must be: uploaded after February 1, 2020, in English, related to e-cigarettes and COVID-19, and less than 30 min in length. We conducted a content analysis of included videos, coding for uploader characteristics, what e-cigarette products were showcased, and specific themes that intersected between e-cigarettes and COVID-19. RESULTS: We examined N = 307 videos and found that N = 220 (73.6%) discussed the health effects of e-cigarette use on COVID-19, followed by videos on how COVID-19 affects e-cigarette sales (N = 40, 12.9%), face mask-related videos (N = 16, 5.1%; e.g., vape tricks including masks) and instructional videos (N = 10, 3.2%; e.g., sanitizing vape devices during COVID-19). Instructional videos had the highest number of likes (Median = 23; IQR = 32) and comments (Median = 10; IQR = 7). CONCLUSIONS: Our findings support the need for continuous surveillance and research on novel vaping-related content in reaction to policies and events, such as the global pandemic. More research is needed to understand the impact of this content on young people's perceptions and use of e-cigarettes.

Understanding e-cigarette content and promotion on YouTube through machine learning.

INTRODUCTION: YouTube is a popular social media used by youth and has electronic cigarette (e-cigarette) content. We used machine learning to identify the content of e-cigarette videos, featured e-cigarette products, video uploaders, and marketing and sales of e-cigarette products. METHODS: We identified e-cigarette content using 18 search terms (eg, e-cig) using fictitious youth viewer profiles and predicted four models using the metadata as the input to supervised machine learning: (1) video themes, (2) featured e-cigarette products, (3) channel type (ie, video uploaders) and (4) discount/sales. We assessed the association between engagement data and the four models. RESULTS: 3830 English videos were included in the supervised machine learning. The most common video theme was 'product review' (48.9%), followed by 'instruction' (eg, 'how to' use/modify e-cigarettes; 17.3%); diverse e-cigarette products were featured; 'vape enthusiasts' most frequently posted e-cigarette videos (54.0%), followed by retailers (20.3%); 43.2% of videos had discount/sales of e-cigarettes; and the most common sales strategy was external links for purchasing (34.1%). 'Vape trick' was the least common theme but had the highest engagement (eg, >2 million views). 'Cannabis' (53.9%) and 'instruction' (49.9%) themes were more likely to have external links for purchasing (p<0.001). The four models achieved an F1 score (a measure of model accuracy) of up to 0.87. DISCUSSION: Our findings indicate that on YouTube videos accessible to youth, a variety of e-cigarette products are featured through diverse videos themes, with discount/sales. The findings highlight the need to regulate the promotion of e-cigarettes on social media platforms.

An Exploration of e-Cigarette-Related Search Items on YouTube: Network Analysis.

BACKGROUND: e-Cigarette use among youth is high, which may be due in part to pro-e-cigarette content on social media such as YouTube. YouTube is also a valuable resource for learning about e-cigarette use, trends, marketing, and e-cigarette user perceptions. However, there is a lack of understanding on how similar e-cigarette-related search items result in similar or relatively mutually exclusive search results. This study uses novel methods to evaluate the relationship between e-cigarette-related search items and results. OBJECTIVE: The aim of this study is to apply network modeling and rule-based classification to characterize the relationships between e-cigarette-related search items on YouTube and gauge the level of importance of each search item as part of an e-cigarette information network on YouTube. METHODS: We used 16 fictitious YouTube profiles to retrieve 4201 distinct videos from 18 keywords related to e-cigarettes. We used network modeling to represent the relationships between the search items. Moreover, we developed a rule-based classification approach to classify videos. We used betweenness centrality (BC) and correlations between nodes (ie, search items) to help us gain knowledge of the underlying structure of the information network. RESULTS: By modeling search items and videos as a network, we observed that broad search items such as e-cig had the most connections to other search items, and specific search items such as cigalike had the least connections. Search items with similar words (eg, vape and vaping) and search items with similar meaning (eg, e-liquid and e-juice) yielded a high degree of connectedness. We also found that each node had 18 (SD 34.8) connections (common videos) on average. BC indicated that general search items such as electronic cigarette and vaping had high importance in the network (BC=0.00836). Our rule-based classification sorted videos into four categories: e-cigarette devices (34%-57%), cannabis vaping (16%-28%), e-liquid (14%-37%), and other (8%-22%). CONCLUSIONS: Our findings indicate that search items on YouTube have unique relationships that vary in strength and importance. Our methods can not only be used to successfully identify the important, overlapping, and unique e-cigarette-related search items but also help determine which search items are more likely to act as a gateway to e-cigarette-related content.

Differences in Health-Related Social Media Usage by Organizations and Individuals.

Introduction: Social media is used as a tool for both information providers and information consumers to disseminate and receive health information. There is a dearth of research that compares the differences between different types of health provider Twitter® (Twitter, Inc., San Francisco, CA) posting styles, specifically regarding the ways in which they communicate health information with the public. This is particularly true for more localized studies that focus on small data sets. Methods: Our study seeks to help fill this gap through an exploration of emergent trends of social media use of small, but specific, stakeholders in Texas, in the United States. Results: A content analysis of health information providers' (individual, organizational, and governmental groups) Tweets based on digital, ethnographic, and grounded theory methods was performed to provide quantitative and qualitative findings in terms of purpose, sentiment, visual features, tone of the Tweets, and public engagement. Conclusions: The findings indicate how individual or organizational users differentially use their Twitter accounts and open up a discussion of what factors might influence effective communication with the public.

Development of a health communication campaign to promote the Texas prescription monitoring program.

Given the role opioid overprescribing has played in the current overdose crisis, reducing the supply of prescription opioids available for misuse has gained widespread support. Prescription monitoring programs (PMPs) have been identified as a tool for achieving this goal, but little is known about how to promote PMP use to prescribers. This paper describes the process of developing a health communication campaign to support the adoption of the Texas PMP. After formative research, message development and concept testing, a range of campaign concepts and messages were tested and final recommendations determined. The messages and lessons learned have utility beyond Texas.

Social media processes in disasters: Implications of emergent technology use.

This article seeks to extend social science scholarship on social media technology use during disruptive events. Though social media's role in times of crisis has been previously studied, much of this work tends to focus on first-responders and relief organizations. However, social media use during disasters tends to be decentralized and this organizational structure can promote different types of messages to top-down information systems. Using 142,786 geo-tagged tweets collected before and after Hurricane Sandy's US landfall as a case study, this article seeks to explore shifts in social media behavior during disruptive events and highlights that though Sandy disrupted routine life within Twitter, users responded to the disaster by employing humor, sharing photos, and checking into locations. We conclude that social media use during disruptive events is complex and understanding these nuanced behaviors is important across the social sciences.

Content Analysis of YouTube Videos Related to E-cigarettes and COVID-19.

INTRODUCTION: E-cigarettes are frequently promoted on social media and portrayed in ways that are attractive to youth. While COVID-19 pandemic significantly affected people's lives, less known is how the pandemic influenced e-cigarette-related marketing and information on social media. This study identifies how e-cigarettes are portrayed during the COVID-19 pandemic on YouTube, one of the most popular social media platforms. METHODS: We searched for combinations of search terms related to e-cigarettes (i.e., "electronic cigarette", "e-cigarette", "e-cig", "vape" and "vaping") and COVID-19 (i.e., "corona", "COVID", "lockdown" and "pandemic"). To be included in the analysis, the video must be: uploaded after February 1, 2020, in English, related to e-cigarettes and COVID-19 and less than 30 minutes in length. We assessed video themes related to e-cigarettes and COVID-19, uploader characteristics, and featured e-cigarette products. RESULTS: We examined N=307 videos and found that N=220 (73.6%) were related to the health effects of e-cigarette use on COVID-19, followed by videos of how COVID-19 affects e-cigarette access/sales (N=40, 12.9%), and face mask-related videos (N=16, 5.1%) which included content regarding masks and e-cigarette use. Instructional videos on how to modify e-cigarettes to use with masks had the highest number of likes (Median=23; IQR=32) and comments (Median=10; IQR=7). CONCLUSIONS: This study identified various e-cigarette contents on YouTube during the COVID-19 pandemic. Our findings support the need for continuous surveillance on novel vaping-related content in reaction to policies and events such as the global pandemic on social media is needed.

Influence of User Profile Attributes on e-Cigarette-Related Searches on YouTube: Machine Learning Clustering and Classification.

Background: The proliferation of e-cigarette content on YouTube is concerning because of its possible effect on youth use behaviors. YouTube has a personalized search and recommendation algorithm that derives attributes from a user's profile, such as age and sex. However, little is known about whether e-cigarette content is shown differently based on user characteristics. Objective: The aim of this study was to understand the influence of age and sex attributes of user profiles on e-cigarette-related YouTube search results. Methods: We created 16 fictitious YouTube profiles with ages of 16 and 24 years, sex (female and male), and ethnicity/race to search for 18 e-cigarette-related search terms. We used unsupervised (k-means clustering and classification) and supervised (graph convolutional network) machine learning and network analysis to characterize the variation in the search results of each profile. We further examined whether user attributes may play a role in e-cigarette-related content exposure by using networks and degree centrality. Results: We analyzed 4201 nonduplicate videos. Our k-means clustering suggested that the videos could be clustered into 3 categories. The graph convolutional network achieved high accuracy (0.72). Videos were classified based on content into 4 categories: product review (49.3%), health information (15.1%), instruction (26.9%), and other (8.5%). Underage users were exposed mostly to instructional videos (37.5%), with some indication that more female 16-year-old profiles were exposed to this content, while young adult age groups (24 years) were exposed mostly to product review videos (39.2%). Conclusions: Our results indicate that demographic attributes factor into YouTube's algorithmic systems in the context of e-cigarette-related queries on YouTube. Specifically, differences in the age and sex attributes of user profiles do result in variance in both the videos presented in YouTube search results as well as in the types of these videos. We find that underage profiles were exposed to e-cigarette content despite YouTube's age-restriction policy that ostensibly prohibits certain e-cigarette content. Greater enforcement of policies to restrict youth access to e-cigarette content is needed.

COVID-19 Misinformation Detection: Machine-Learned Solutions to the Infodemic.

Background: The volume of COVID-19-related misinformation has long exceeded the resources available to fact checkers to effectively mitigate its ill effects. Automated and web-based approaches can provide effective deterrents to online misinformation. Machine learning-based methods have achieved robust performance on text classification tasks, including potentially low-quality-news credibility assessment. Despite the progress of initial, rapid interventions, the enormity of COVID-19-related misinformation continues to overwhelm fact checkers. Therefore, improvement in automated and machine-learned methods for an infodemic response is urgently needed. Objective: The aim of this study was to achieve improvement in automated and machine-learned methods for an infodemic response. Methods: We evaluated three strategies for training a machine-learning model to determine the highest model performance: (1) COVID-19-related fact-checked data only, (2) general fact-checked data only, and (3) combined COVID-19 and general fact-checked data. We created two COVID-19-related misinformation data sets from fact-checked "false" content combined with programmatically retrieved "true" content. The first set contained ~7000 entries from July to August 2020, and the second contained ~31,000 entries from January 2020 to June 2022. We crowdsourced 31,441 votes to human label the first data set. Results: The models achieved an accuracy of 96.55% and 94.56% on the first and second external validation data set, respectively. Our best-performing model was developed using COVID-19-specific content. We were able to successfully develop combined models that outperformed human votes of misinformation. Specifically, when we blended our model predictions with human votes, the highest accuracy we achieved on the first external validation data set was 99.1%. When we considered outputs where the machine-learning model agreed with human votes, we achieved accuracies up to 98.59% on the first validation data set. This outperformed human votes alone with an accuracy of only 73%. Conclusions: External validation accuracies of 96.55% and 94.56% are evidence that machine learning can produce superior results for the difficult task of classifying the veracity of COVID-19 content. Pretrained language models performed best when fine-tuned on a topic-specific data set, while other models achieved their best accuracy when fine-tuned on a combination of topic-specific and general-topic data sets. Crucially, our study found that blended models, trained/fine-tuned on general-topic content with crowdsourced data, improved our models' accuracies up to 99.7%. The successful use of crowdsourced data can increase the accuracy of models in situations when expert-labeled data are scarce. The 98.59% accuracy on a "high-confidence" subsection comprised of machine-learned and human labels suggests that crowdsourced votes can optimize machine-learned labels to improve accuracy above human-only levels. These results support the utility of supervised machine learning to deter and combat future health-related disinformation.

CoVerifi: A COVID-19 news verification system.

There is an abundance of misinformation, disinformation, and "fake news" related to COVID-19, leading the director-general of the World Health Organization to term this an 'infodemic'. Given the high volume of COVID-19 content on the Internet, many find it difficult to evaluate veracity. Vulnerable and marginalized groups are being misinformed and subject to high levels of stress. Riots and panic buying have also taken place due to "fake news". However, individual research-led websites can make a major difference in terms of providing accurate information. For example, the Johns Hopkins Coronavirus Resource Center website has over 81 million entries linked to it on Google. With the outbreak of COVID-19 and the knowledge that deceptive news has the potential to measurably affect the beliefs of the public, new strategies are needed to prevent the spread of misinformation. This study seeks to make a timely intervention to the information landscape through a COVID-19 "fake news", misinformation, and disinformation website. In this article, we introduce CoVerifi, a web application which combines both the power of machine learning and the power of human feedback to assess the credibility of news. By allowing users the ability to "vote" on news content, the CoVerifi platform will allow us to release labelled data as open source, which will enable further research on preventing the spread of COVID-19-related misinformation. We discuss the development of CoVerifi and the potential utility of deploying the system at scale for combating the COVID-19 "infodemic".

Who tweets about cancer? An analysis of cancer-related tweets in the USA.

Cancer patients, family members and friends are increasingly using social media. Some oncologists and oncology centres are engaging with social media, and advocacy groups are using it to disseminate information and coordinate fundraising efforts. However, the question of whether such social media activity corresponds to areas with higher incidence of cancer or higher access to cancer centres remains understudied. To address this gap, our study compared US government data with 90,986 cancer-related tweets with the keywords 'chemo', 'lymphoma', 'mammogram', 'melanoma', and 'cancer survivor'. We found that the frequency of cancer-related tweets is not associated with mammogram testing and cancer incidence rates, but that the concentration of doctors and cancer centres is associated with cancer-related tweet frequency. Ultimately, we found that Twitter has value to cancer patients, survivors and their families, but that cancer-related social media resources may not be targeting locations that could see the most value and benefit. Therefore, there are real opportunities to better align cancer-related engagement on Twitter and other social media.

Modeling virtual organizations with Latent Dirichlet Allocation: a case for natural language processing.

This paper explores a variety of methods for applying the Latent Dirichlet Allocation (LDA) automated topic modeling algorithm to the modeling of the structure and behavior of virtual organizations found within modern social media and social networking environments. As the field of Big Data reveals, an increase in the scale of social data available presents new challenges which are not tackled by merely scaling up hardware and software. Rather, they necessitate new methods and, indeed, new areas of expertise. Natural language processing provides one such method. This paper applies LDA to the study of scientific virtual organizations whose members employ social technologies. Because of the vast data footprint in these virtual platforms, we found that natural language processing was needed to 'unlock' and render visible latent, previously unseen conversational connections across large textual corpora (spanning profiles, discussion threads, forums, and other social media incarnations). We introduce variants of LDA and ultimately make the argument that natural language processing is a critical interdisciplinary methodology to make better sense of social 'Big Data' and we were able to successfully model nested discussion topics from forums and blog posts using LDA. Importantly, we found that LDA can move us beyond the state-of-the-art in conventional Social Network Analysis techniques.