Myopia control effect of orthokeratology lens and construction and evaluation of prediction model based on random forest and Logistic regression
10.13389/j.cnki.rao.2024.0122
- VernacularTitle:基于随机森林、Logistic回归的角膜塑形镜近视控制效果及预测模型构建和评价
- Author:
Hua ZHANG
1
;
Shaofang GAO
;
Xin ZHAO
;
Jinyuan MI
Author Information
1. 050030 河北省石家庄市,石家庄市人民医院
- Keywords:
orthokeratology lens;
random forest;
prediction model;
myopia;
control effect
- From:
Recent Advances in Ophthalmology
2024;44(8):637-642
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore the myopia control effect of orthokeratology(OK)lenses and build a prediction model of myopia control effect based on random forest and Logistic regression,so as to provide a basis for further impro-ving adolescent vision.Methods A total of 289 patients who wore OK lenses after seeing an ophthalmologist in Shiji-azhuang People's Hospital from January to June 2019 were selected as the OK lens group,and a total of 289 patients who wore framed glasses during the same period were selected as the framed glass group.After 2 and 4 years of wearing,the in-crease in axial length(AL)of the patients was observed.The patients were divided into a well-controlled myopia group and a poorly-controlled myopia group according to the changes in AL of the eyes after wearing OK lenses for 4 years.The fac-tors influencing the myopia control effect were analyzed based on random forest and Logistic regression,and a nomogram prediction model was constructed.The performance of the model was evaluated using the receiver operating characteristic(ROC)curve,calibration curve,and decision curve analysis(DCA).Results After 2 and 4 years,the diopter increase and AL increase of patients wearing OK lenses were lower than those of patients wearing framed glasses(all P＜0.001).There were no significant changes in corneal endothelial cell parameters and central corneal thickness at the thinnest point before,2 years and 4 years after wearing between the two groups(all P＞0.05).The random forest model showed that basic AL,daily wear time,correct eye health behavior,pupil diameter,daily outdoor activity time,homework time after school,and daily sleep time were the top 7 variables in importance ranking.Logistic regression analysis showed that home-work time after school was a risk factor for poor myopia control in OK lens wearers,while basic AL,pupil diameter,daily wear time,correct eye health behavior,daily outdoor activity time,and daily sleep time were protective factors for myopia control.The ROC curve,calibration curve,and DCA suggested that the nomogram prediction model constructed based on random forest and Logistic regression analysis had good accuracy,consistency and clinical effectiveness.Conclusion OK lens shows good myopia control effect,has little effect on corneal endothelial cells and corneal thickness,and demon-strates high safety.Homework time after school,basic AL,pupil diameter,daily wear time,correct eye health behavior,daily outdoor activity time,and daily sleep time are influencing factors of myopia control.Strengthening the construction of a prediction model is helpful in identifying high-risk groups with poor myopia control and guiding clinical interventions in time.
