Development of A Myopia Occurrence Risk Prediction Nomogram Model Based on LASSO-logistic Regression

Xie Shuqian; Yang Zhengwei; Qian Wenjun; Chen Kai

doi:10.3969/j.issn.1001-568X.2025.10.0014

Abstract：OBJECTIVE To investigate the prevalence of myopia among students in Yixing of Jiangsu Province,analyze influencing factors,and develop a logistic regression-based nomogram prediction model,thereby providing scientific support for establishing a precision intervention system for adolescent vision health in Yixing.METHODS Under the framework of the 2024 Jiangsu Provincial Student Common Diseases and Health Influencing Factors Surveillance Program,vision screening and questionnaire surveys were conducted among 1 863 students in Yixing.A multi-stage statistical modeling strategy was employed:variables were preliminarily screened using χ2 tests(α=0.05),followed by LASSO regression for feature selection,and finally a logistic regression model was constructed.The nomogram prediction model for myopia risk was developed using the rms package in R 4.4.2.Model accuracy was evaluated through Receiver Operating Characteristic(ROC)curves and Calibration curves.RESULTS Among 1 863 students,1 630(87.49%)had myopia.Logistic regression revealed significant risk factors:junior high school,senior high school,vocational high school,one myopic parent,both myopic parents,and extracurricular tutoring starting at≥6 years.The nomogram model incorporating educational stage,parental myopia,and tutoring initiation age achieved an AUC of 0.773(95%CI:0.741～0.806),with calibration curves confirming good predictive performance.CONCLUSION Myopia prevalence escalates with advancing education stages and correlates with parental myopia and extracurricular tutoring.The developed model demonstrates robust accuracy and discriminative ability,enabling quantified myopia risk assessment via nomogram visualization.