中国幼儿近视预测模型的开发与验证

Development and validation of a model for predicting myopia in young children in China.

作者信息

Hu Ting, Wu Rong, He Xi, Hua Zi-Yu, Peng Tian-Li, Peng Xiao-Yan, Li Shi-Ming

机构信息

Beijing Yanqing District Hospital (Peking University Third Hospital Yanqing Hospital), Yanqing District, Beijing, China.

Department of Ophthalmology, Beijing Tongren Hospital, Capital Medical University, Beijing, China.

出版信息

BMC Ophthalmol. 2025 Jul 31;25(1):439. doi: 10.1186/s12886-025-04267-6.

DOI:10.1186/s12886-025-04267-6

PMID:40745285

原文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC12312385/

Abstract

BACKGROUND

Early detection of myopia risk and timely intervention are key to reducing the incidence of myopia. This study aimed to construct a myopia prediction model that included perinatal factors and evaluate the correlation between each factor and myopia.

METHODS

A cross-sectional study was conducted. A questionnaire was used to investigate the related factors. The predictors of myopia onset were determined via univariate analysis (Model A), LASSO regression and 10-fold cross validation (Model B). The perinatal factors in Model B were removed to form Model C. Binary logistic regression was used to establish the model. Then, internal and external verification were performed. The differentiation, calibration, and clinical benefits of the three models were compared.

RESULTS

A total of 1322 participants with an average age of 7.85 ± 1.58 years were included in the study. Model B included age, axial length (AL), AL to corneal curvature (AL/CR) ratio, steep K (K2) value, parental myopia, and maternal history of gestational hypertension. Neither Model A nor Model C included perinatal factors. Model B had the best discrimination ability in the development, internal validation, and external validation groups and showed the best consistency and clinical benefit in all three groups. After model standardization, the effects of the respective variables on the results were as follows: AL/CR ratio: 59.14, maternal history of gestational hypertension: 44.03, AL: 30.47, K2: 5.65, parental myopia: 4.55, and age: 0.35.

CONCLUSIONS

A maternal history of gestational hypertension was associated with myopia. The effect of perinatal factors on myopia deserves attention.

TRIAL REGISTRATION

ChiCTR2200065398. November 3, 2022. (http://www.chictr.org.cn).

摘要

背景

早期发现近视风险并及时干预是降低近视发病率的关键。本研究旨在构建一个包含围产期因素的近视预测模型，并评估各因素与近视之间的相关性。

方法

进行了一项横断面研究。采用问卷调查相关因素。通过单因素分析（模型A）、LASSO回归和10倍交叉验证（模型B）确定近视发病的预测因素。去除模型B中的围产期因素以形成模型C。采用二元逻辑回归建立模型。然后进行内部和外部验证。比较了三种模型的区分度、校准度和临床效益。

结果

本研究共纳入1322名参与者，平均年龄为7.85±1.58岁。模型B包括年龄、眼轴长度（AL）、眼轴与角膜曲率（AL/CR）比值、角膜陡峭曲率（K2）值、父母近视以及母亲妊娠期高血压病史。模型A和模型C均未包括围产期因素。模型B在发育组、内部验证组和外部验证组中具有最佳的区分能力，并且在所有三组中均显示出最佳的一致性和临床效益。模型标准化后，各变量对结果的影响如下：AL/CR比值：59.14，母亲妊娠期高血压病史：44.03，AL：30.47，K2：5.65，父母近视：4.55，年龄：0.35。