A Machine Learning Approach to Explore Perceived Stress in College Students Post COVID-19 Pandemic

The perceived stress among college students post COVID-19 pandemic has been a major concern for health researchers. Machine learning (ML) techniques were developed to identify college students at risk of perceived stress. The dataset included 291 Chinese university students. Perceived stress was measured by 10-item Perceived Stress Scale (PSS-10). Features were first selected through LASSO regression. Then six ML prediction models were built, and their performances were evaluated. The Shapley Additive exPlanations (SHAP) model was used to explain the results. The Extreme Gradient Boosting (XGBoost) model demonstrated the best overall performance, achieving an area under the curve (AUC) of 0.788. Academic pressure, health, insomnia, gender, and catch-up were identified as the primary predictors of significant perceived stress among college students post COVID-19 pandemic. This study offers a practical tool and valuable insights for health professionals in the public sector to identify students experiencing high levels of perceived stress.