A Review on the Recent Advancements in Machine Learning-Assisted Tobacco Research

Abstract

Tobacco smoking, a highly complex behavior influenced by genetic predisposition and environmental factors, is a grave global health alarm and a leading preventable cause of death, linked to severe diseases like osteoporosis and various cancers. Quitting rates remain dishearteningly low despite recommendations for nicotine replacement therapy and healthcare provider discussions. Presently, tobacco research generates vast data that can be harnessed by machine learning models, including supervised, unsupervised, and deep learning algorithms, which are gaining traction in tobacco research. This review delves into the intersection of traditional tobacco research and machine learning, elucidating the potential of ML methodologies in addressing the challenges of tobacco control. By leveraging diverse applications such as identifying smoking susceptibility predictors, predicting smoking behavior from genetic data, aiding cessation efforts with personalized predictors, automating data collection through apps, managing nicotine cravings, tracking smoking-induced diseases, monitoring illegal online vape sales, and assessing second-hand and third-hand smoke exposure, especially in infants, ML emerges as a powerful tool to combat the tobacco epidemic. Furthermore, this review highlights the significance of integrating ML techniques in tobacco research, emphasizing their role in advancing our understanding of smoking behavior, informing targeted interventions, and ultimately mitigating the public health burden associated with tobacco use. By harnessing the predictive power of ML algorithms, researchers and policymakers can tailor interventions to individual needs, optimize resource allocation, and accelerate progress toward a tobacco-free future.