Identifying Key Predictors of Smoking Cessation Success: Text-Based Feature Selection Using a Large Language Model

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: AUC 0.752 for GPT-4.1 selected variables vs 0.749 for all variables

• Primary Outcome: 12-month smoking abstinence at wave 6 follow-up

• Nnt Or Sensitivity: Top predictors include smoking frequency, time to first cigarette, social influences, and emotional dependence

• Confidence Interval: Not reported

Clinical Application

High - identified predictors are readily assessable in clinical practice through brief questionnaires Relevant to Australian practice; predictors align with RACGP smoking cessation guidelines. Could inform PBS-subsidised cessation programs and Quitline services Adult smokers attempting cessation, particularly in primary care and specialist smoking cessation services

Abstract

INTRODUCTION: The most effective way to reduce mortality and morbidity among current smokers is to quit smoking. Although about half of smokers attempted to quit, only one-tenth succeeded in 2022. Understanding key predictors of smoking cessation success would inform smoking cessation interventions and increase quitting rates. METHODS: We analyzed data from waves 5 and 6 of the Population Assessment of Tobacco and Health (PATH) study (December 2018 to November 2021). Using OpenAI's GPT-4.1, we identified the top 45 variables from wave 5 that are highly predictive of 12-month smoking abstinence in wave 6, based on descriptions of survey variables. We then validated the predictive power of the GPT-4.1-selected variables by comparing the performance of eXtreme Gradient Boosting (XGBoost) trained on different sets of variables. Finally, we derived insights into the top 10 variables, ranked according to their SHapley Additive exPlanations values. RESULTS: The performance of XGBoost trained with all possible wave 5 variables and the 45 selected variables was almost identical (AUC:0.749 vs AUC:0.752). The top 10 variables included past 30-day smoking frequency, minutes from waking up to smoking first cigarette, important people's views on tobacco use, prevalence of tobacco use among close associates, daily electronic nicotine product use, emotional dependence, and health harm concerns. CONCLUSIONS: The high predictive performance of XGBoost, when trained on the selected variables, underscores the efficiency and efficacy of GPT-4.1-based feature selection. The top 10 variables include various risk factors that have been previously reported in the literature for their influence on smoking behavior. IMPLICATIONS: Our findings do not establish causal relationships between the selected predictors and 12-month smoking abstinence. However, identifying these key predictors provides valuable insights into the factors highly associated with smoking cessation success. This study demonstrates the ability of OpenAI's GPT-4.1 to perform feature selection using only the textual descriptions of variables. The efficient and successful application of GPT-4.1 for variable selection highlights the potential of integrating artificial intelligence tools into tobacco research to guide resource-efficient and targeted intervention strategies.

References

1. 1.Le, T. T. T., Yang, J., Zhao, Z., Zhang, K., Li, W., & Hu, Y. (2026). Identifying Key Predictors of Smoking Cessation Success: Text-Based Feature Selection Using a Large Language Model. Nicotine & Tobacco Research. https://doi.org/10.1016/j.drugalcdep.2017.05.010