Exploring Factors Associated with Physical Exercise Participation Among Chinese Adults Based on Explainable Machine...

Area of Science:

• Public Health
• Behavioral Science
• Data Science

Background:

• Physical inactivity is a significant public health issue in China, with only 30.3% of adults engaging in regular exercise.
• Understanding factors influencing exercise participation is crucial for developing effective health promotion strategies.

Purpose of the Study:

• To identify key predictors of physical exercise participation in China.
• To explore social, psychological, and environmental factors influencing exercise behavior.

Main Methods:

• Utilized 2021 China General Social Survey (CGSS) data.
• Employed dimensionality reduction, machine learning (SVM), and SHAP interpretability analysis.
• Applied LassoCV for feature selection among 19 potential factors.

Main Results:

• Support Vector Machine (SVM) model demonstrated superior predictive accuracy.
• Key predictors identified: sports viewing, household registration, education, well-being, smoking, age, sleep quality, social activities, and residence suitability.
• Positive associations found with higher education, well-being, urban residency, sports viewing, and suitable residence; negative associations with smoking and poor sleep.

Conclusions:

• Machine learning combined with interpretability methods effectively identifies exercise behavior predictors.
• Findings offer novel insights into social, psychological, and environmental determinants of physical activity in China.
• Results can inform targeted interventions to boost exercise participation and improve public health.