The impact of physical exercise on adolescent psychological resilience: an empirical analysis based on deep neural networks

Abstract

Objective: The junior high school years represent a crucial period in adolescents’ physical and psychological development, during which mental health concerns have gained significant attention. Psychological resilience, a modifiable protective factor, is vital for maintaining mental wellbeing. Method: This study involved 1,064 junior high school students, who were selected through stratified random sampling. Data were collected using the Physical Activity Rating Scale-3 (PARS-3) and the Adolescent Resilience Scale. The analysis integrated multiple linear regression, deep neural networks (DNN), and the SHAP (Shapley Additive Explanations) model for interpretability. Results: Physical exercise intensity, duration, and frequency were significantly positively correlated with adolescents’ psychological resilience and its subdimensions. Notably, the interaction between intensity and duration was prominent, exhibiting nonlinear effects. Furthermore, the interaction between grade level and physical exercise was statistically significant, highlighting lower grades as a key intervention period. For individuals with low psychological resilience, increasing both exercise intensity and duration led to more effective intervention outcomes. Conclusion: Physical exercise is significantly positively associated with adolescent psychological resilience, and this association may operates through potential nonlinear pathways. Furthermore, exercise characteristics may interact with demographic variables such as grade level, providing empirical insights to inform the development of tailored intervention strategies.