基于回归模型的青少年裸眼视力与验光指标相关性分析研究  

Analysis of the correlation between uncorrected visual acuity and refractive parameters of adolescents based on regression model

作　　者：叶汪洋 张晨曦 丁柯军 金恺睿 魏玮[2] 诸科璇 李晔锋[1] Ye Wang-yang;Zhang Chen-xi;Ding Ke-jun(Ningbo University of technology,Ningbo 315211,China)

机构地区：[1]宁波工程学院,315211 [2]宁波市医疗中心李惠利医院

出　　处：《中国斜视与小儿眼科杂志》2022年第1期8-10,4,共4页Chinese Journal of Strabismus & Pediatric Ophthalmology

基　　金：浙江省基础公益研究计划项目(编号:LGF22H120009)。

摘　　要：目的研究7~13岁儿童的验光指标与裸眼视力之间的相关度,以及尝试把验光指标拟合成裸眼视力值并评估其性能。方法首先根据Pearson分析找到验光指标与裸眼视力之间的相关度,之后使用四种回归算法模型以及两个复合模型拟合裸眼视力,使用R;得分与RMSE评估模型的优劣。结果球镜对于裸眼视力的相关度最高,眼轴长度次之;支持向量回归、梯度提升回归模型以及均值、Stacking复合模型对裸眼视力有较好的拟合性(>0.85)。结论回归模型能够较为准确地拟合学龄儿童的裸眼视力,并可能有助于学校视力普查。Objective To investigate the correlation between refractive parameters and uncorrected visual acuity in children aged 7~13 years,and fit the refractive parameters to the uncorrected visual acuity value along with the evaluation of the performance.Method It firstly used pearson analysis to work out the correlation between refractive parameters and uncorrected visual acuity value,and then it applied four regression algorithm models and two composite models to fit the uncorrected visual acuity value,with R;scores and RMSE values for evaluation.Result SPH had the highest correlation for uncorrected visual acuity,followed by AXL;Single models such as support vector regression,gradient boost regression,and composite models such as mean and stacking had a better fit(> 0.85) in fitting the uncorrected visual acuity.Conclusion The regression models were able to fit the uncorrected visual acuity of school-age children more accurately and may be useful for school vision screening.

关 键 词：裸眼视力 验光指标 回归模型 相关性分析 

分 类 号：R778.2[医药卫生—眼科]