Prediction of myopia progression in children wearing highly aspherical lenslet spectacles

This study aims to examine the 1-year changes in spherical equivalent (SE) refraction and ocular biometric parameters among Chinese children wearing highly aspherical lenslet spectacle and develop a predictive model for identifying children at risk of significant myopic progression (≥0.50 Diopter).

In this retrospective case-control study, children who started wearing highly aspherical lenslet spectacles for 1 year were enrolled and randomly divided into training and validation sets. Data on SE refraction and ocular biometric parameters were collected at baseline, 6 months, and 12 months. A logistic regression (LR) model was developed and compared with six machine learning models. Model performance was assessed by the area under the receiver operating characteristic curve, calibration, and decision curve analysis (DCA).

Over 12 months, the rapid and slow progression groups displayed divergent trends in their SE refraction and ocular biometric parameters. Baseline age and axial length were the only significant predictors in the multivariate analysis. The extreme gradient boosting model achieved the highest discriminative performance. The LR model demonstrated good discrimination, acceptable calibration, and clinical utility across a broad range of threshold probabilities in DCA. A nomogram was developed from the LR model to visualize individual risk.

This study developed practical predictive models to identify children at high risk of rapid myopia progression while wearing highly aspherical lenslet spectacles. These findings provide a clinically applicable, evidence-based tool for personalized myopia management, supporting early stratification and tailored intervention to improve myopia control outcomes in children. However, these results are population-specific and should be cautiously generalized to pediatric cohorts from other ethnic groups or regions.