Development and Internal Validation of a Predictive Model of Perceived Stress Among Military Students: A LASSO...

This study aimed to develop and internally validate a predictive model of perceived stress among first-year military male students to examine the predictive contribution of personality traits, depressive symptoms, and psychological well-being. Understanding these psychological predictors may support interventions for students at elevated risk of stress during military and academic transition. A cross-sectional web-based survey included 274 first-year male students at the Military Technological College in Oman. Outcome measures included the Perceived Stress Scale (PSS-10), the Patient Health Questionnaire (PHQ-9) for depressive symptoms, the WHO-5 Well-being Index, and the Big Five Inventory assessing personality traits. All variables were analyzed as continuous measures. Predictive modeling was performed using Least Absolute Shrinkage and Selection Operator (LASSO) linear regression with repeated 70/30 train-test splitting across 100 iterations and 10-fold cross-validation for internal validation. The final analytic sample included 266 participants after exclusion of incomplete responses. Across the 100 internal validation runs, the LASSO model accounted for approximately 40% of the variance in perceived stress (training R² = 0.44 ± 0.04; test R² = 0.40 ± 0.08). Neuroticism (β = 0.35) and depressive symptoms (β = 0.15) showed positive associations with perceived stress, whereas psychological well-being showed a negative association (β = -0.32). PHQ-9, WHO-5, and neuroticism were selected in 100% of the repeated LASSO models, which showed the most stable predictive contribution. Model performance on the test datasets showed stable predictive accuracy (MSE = 20.24 ± 2.48; RMSE = 4.49 ± 0.28; MAE = 3.61 ± 0.23). These findings demonstrate that personality traits, depressive symptoms, and psychological well-being collectively contribute to the statistical modeling of perceived stress among military students. The internally validated associative model may support institutional interventions for students vulnerable to elevated stress, informing targeted preventive mental health strategies within military training environments.