Analyzing adolescents’ e-cigarette initiation tendency using explainable machine learning

Napat Seelpipat and Daricha Sutivong ()
Additional contact information
Napat Seelpipat: Chulalongkorn University
Daricha Sutivong: Chulalongkorn University

Journal of Computational Social Science, 2025, vol. 8, issue 4, No 14, 39 pages

Abstract: Abstract E-cigarette curiosity and susceptibility serve as key indicators of future e-cigarette initiation tendency. Data from the 2019 (pre-COVID-19) and 2022 (post-COVID-19) National Youth Tobacco Surveys were analyzed to holistically identify and rank influencing factors using machine learning techniques. To examine temporal shifts, we compared top predictors from both years using common features, and contrasted them with those derived directly from the full 2022 dataset. Factors from survey responses were categorized into background information, tobacco product experience, perceptions and attitudes toward tobacco products, exposure to advertisements, etc. Among the explored three tree-based algorithms, Random Forest (RF) consistently demonstrated superior performance in both years. The SHapley Additive exPlanations (SHAP) analysis reveals that perceptions and attitudes toward tobacco products remained the most influential attributes before and after COVID-19. Specifically, strongly agreeing that all tobacco products are dangerous and perceiving greater harm from occasional e-cigarette use were key to reducing curiosity and susceptibility. However, after COVID-19, social media-related and peer-related influences overtook the previously dominant role of exposure to e-cigarettes in public or school settings. By 2022, social media interactions with e-cigarette content, along with perceptions of peer acceptability and estimates of peer usage, emerged as significant factors—surpassing secondhand smoke and vapor exposure. Furthermore, in 2022, emotional well-being rose in its relevance; students reporting negative emotional symptoms exhibited an increased likelihood of curiosity and susceptibility. These findings can inform regulatory and administrative agencies in developing more targeted preventive interventions and guide future survey designs on exploring the evolving factors influencing adolescent e-cigarette use.

Keywords: E-cigarette use; National Youth Tobacco Survey (NYTS); Machine learning; Random Forest (RF); Extreme Gradient Boosting (XGBoost); Shapley Additive Explanations (SHAP) (search for similar items in EconPapers)
Date: 2025
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://link.springer.com/10.1007/s42001-025-00423-6 Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:jcsosc:v:8:y:2025:i:4:d:10.1007_s42001-025-00423-6

Ordering information: This journal article can be ordered from
http://www.springer. ... iences/journal/42001

DOI: 10.1007/s42001-025-00423-6

Access Statistics for this article

Journal of Computational Social Science is currently edited by Takashi Kamihigashi

More articles in Journal of Computational Social Science from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().