Determinants of Refractive Error Among School-Going Children in North India: A Machine Learning-Enhanced Analysis Using Elastic Net Regression

Abstract: Background: Refractive error is a leading cause of visual impairment among school-aged children globally, yet population-specific determinants remain inadequately characterized in resource-limited settings. Objective: This study sought to quantify the prevalence of refractive error and develop predictive models identifying sociodemographic, environmental, dietary, and behavioral determinants among school-going children in Mathura, Uttar Pradesh. Methods: A cross-sectional study was conducted among 2,000 school children aged 6-16 years. Comprehensive vision screening was performed, and data on sociodemographic characteristics, housing conditions, dietary patterns, lifestyle factors, and academic variables were collected through structured questionnaires. Multivariable logistic regression and elastic net regularized regression were used to identify independent predictors of refractive error. Results: The prevalence of refractive error was 24.1% (482/2,000). In multivariable analysis, urban residence (adjusted OR 1.71, 95% CI: 1.31-2.22, p