Examination of social smoking classifications using a machine learning approach.

Area of Science:

• Behavioral Science
• Psychology
• Public Health

Background:

• Social smoking definitions vary, hindering research comparability.
• Understanding social smoking is crucial for targeted interventions.

Purpose of the Study:

• To investigate and differentiate four distinct social smoking classifications.
• To utilize traditional and machine learning models for classification.

Main Methods:

• 133 young adults completed self-report measures and a simulated social smoking context assessment.
• A multilayer perceptron artificial network was employed as a novel machine learning approach.

Main Results:

• Three of four social smoking definitions showed lower dependence, frequency, quantity, and willingness to smoke compared to non-social smokers.
• Machine learning successfully differentiated all four classifications.
• Past month smoking frequency was the most critical predictor across models.

Conclusions:

• Social smoking definitions exhibit significant variability in user characteristics.
• Machine learning offers a robust method for classifying social smokers.
• Recommendations are provided for selecting appropriate social smoker classifications in future studies.