SHAP-Assisted XGBoost Analysis of Psychological Flourishing in Deaf University Students: Contributions of Identity Integration, Self-Compassion, and Perceived Inclusion

Abstract

The present study aimed to employ a SHAP-assisted Extreme Gradient Boosting (XGBoost) model to predict psychological flourishing among deaf university students and determine the relative contributions of identity integration, self-compassion, and perceived inclusion to flourishing outcomes. This cross-sectional predictive study was conducted among 428 deaf university students enrolled in Canadian higher education institutions during the 2025–2026 academic year. Participants completed validated measures of psychological flourishing, identity integration, self-compassion, and perceived inclusion. Data were analyzed using XGBoost machine learning algorithms combined with Shapley Additive Explanations (SHAP) to identify the relative importance of predictor variables and provide interpretable explanations of model outputs. The dataset was randomly divided into training and testing subsets using an 80:20 ratio. Five-fold cross-validation and hyperparameter optimization procedures were employed to maximize predictive accuracy and model generalizability. Model performance was evaluated using the coefficient of determination (R²), root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). The XGBoost model demonstrated excellent predictive performance, explaining 87.2% of the variance in psychological flourishing in the training dataset and 82.4% in the testing dataset. SHAP analyses identified perceived inclusion as the most influential predictor, accounting for 39.8% of total model importance, followed by identity integration (33.6%) and self-compassion (22.6%). Correlation analyses further revealed significant positive associations between flourishing and perceived inclusion (r = .76, p < .001), identity integration (r = .71, p < .001), and self-compassion (r = .65, p < .001). SHAP dependence patterns indicated that higher levels of all three predictors were associated with substantial increases in predicted flourishing scores, with perceived inclusion exhibiting the strongest and most consistent positive effects. The findings demonstrate that psychological flourishing among deaf university students can be predicted with high accuracy using explainable machine learning approaches. Perceived inclusion emerged as the strongest determinant of flourishing, followed by identity integration and self-compassion, highlighting the critical importance of inclusive educational environments and positive self-related psychological resources. The study underscores the value of SHAP-assisted XGBoost models for advancing understanding of well-being among marginalized student populations and provides evidence for developing interventions that strengthen inclusion, identity affirmation, and self-compassion within higher education settings.