A Psychometric Framework for Modeling the impact of Insomnia on Academic Performance

Insomnia is increasingly prevalent among university students, adversely affecting both academic performance and psychological well being. This study presents a comprehensive behavioral modeling framework linking sleep disruption to academic outcomes using self reported psychometric data from 996 undergraduate students across all four academic years. The dataset captured sleep patterns, lifestyle factors, stress, and academic performance indicators through a structured questionnaire survey. A significant negative correlation was observed between sleep disturbances and GPA. Unsupervised clustering of sleep and lifestyle features revealed distinct behavioral subgroups, while supervised classifiers validated through both 80/20 train–test split and 3-fold cross-validation achieved strong predictive performance (AUC = 0.96, F1 = 0.89). Permutation based feature importance identified sleep-onset difficulty, perceived stress, and caffeine consumption as dominant predictors of academic decline. These findings demonstrate the value of integrating psychometric assessment with machine learning to identify students at academic risk and provide a foundation for data-driven interventions in higher education. • Sleep disturbance strongly predicts lower student academic outcomes • Machine learning achieved AUC = 0.96, F1 = 0.89 in performance prediction • 93% of students met DSM-based criteria for insomnia-like symptoms. • Exercise moderated the negative effect of poor sleep on GPA. • Stress lowers GPA; caffeine’s impact depends on time of intake.