Development and validation of a risk prediction model for myopia onset in children and adolescents

This study aimed to identify the key risk factors for pediatric myopia and construct a validated predictive model to achieve precise myopia risk assessment in children and adolescents. A systematic review and meta-analysis based on five major databases were performed to screen and determine the risk factors of pediatric myopia. Statistically significant factors ( P < 0.05) were weighted by pooled odds ratios (ORs) and β coefficients to establish a logistic regression prediction model and corresponding nomogram. A total of 4,045 participants from a prospective cohort were enrolled for model validation, with receiver operating characteristic (ROC), area under the curve (AUC), calibration curves, and decision curve analysis (DCA) adopted to evaluate the model’s predictive efficacy, calibration degree, and clinical application value. Data from 12 study cohorts involving 81,723 participants were included in the final analysis. Six independent risk factors for pediatric myopia were identified: girl (OR=1.39, 95% CI 1.11–1.74), older age (OR=1.15, 95% CI 1.11–1.19), parental myopia (one myopic parent: OR=1.44, 95% CI 1.26–1.64; two myopic parents: OR=1.92, 95% CI 1.48–2.49), longer axial length (OR=3.18, 95% CI 1.37–7.39), insufficient outdoor activity time (OR=0.60, 95% CI 0.44–0.82), and prolonged near-work time (OR=1.05, 95% CI 1.03–1.08). The AUC values of the basic model and the optimized extended model were 0.771 (95% CI 0.756–0.786) and 0.789 (95% CI 0.775–0.803), respectively. The calibration slope of the extended model was closer to the ideal value of 1 than that of the base model. DCA verified excellent clinical utility of the model within the threshold probability range of 0.20–0.90. Risk stratification analysis based on the established model showed significant differences in myopia prevalence across subgroups: 80.3% in the very high-risk group, 42.2% in the high-risk group, 24.4% in the moderate-risk group, and 19.1% in the low-risk group (χ 2 =866.84, P < 0.001). This study constructed a robust pediatric myopia risk prediction model based on meta-analysis evidence with reliable predictive performance. The validated model can provide a scientific and quantitative basis for individualized early risk warning and tiered targeted intervention strategies for childhood and adolescent myopia.