应用Logistic回归和ROC曲线评价患者年龄对儿童子午线性弱视诊断的影响

目的：应用Logistic多元回归分析和ROC曲线探讨年龄因素对诊断儿童子午线弱视有无影响。方法：研究对象为2008／2011年间在我院眼科门诊就诊，以散光为主要屈光异常并排除屈光参差及斜视的4～8岁儿童共1005例1910眼。采用Logistic多元回归分析年龄、性别、柱镜绝对值程度、球镜绝对值程度、散光类型对诊断子午线性弱视的影响，通过ROC曲线下面积（areaundertheROCcurve，AUC）分析进一步明确患者年龄因素对诊断子午线弱视的影响。结果：分别建立Logistic回归模型1（包括性别、柱镜绝对值程度、球镜绝对值程度、散光类型四个变量）和模型2（前四个变量再加上年龄）。两个...

Full description

Saved in:
Bibliographic Details
Published in国际眼科杂志 Vol. 13; no. 7; pp. 1434 - 1437
Main Author 李辉 许江涛 蒋晓明 周莹
Format Journal Article
LanguageChinese
Published 中国云南省昆明市儿童医院眼科,650034 2013
Subjects
Logistic模型
ROC
儿童
子午线性弱视
散光
子午线性弱视
散光
儿童
Logistic模型
ROC
Online AccessGet full text

Cover

More Information
Summary:目的：应用Logistic多元回归分析和ROC曲线探讨年龄因素对诊断儿童子午线弱视有无影响。方法：研究对象为2008／2011年间在我院眼科门诊就诊，以散光为主要屈光异常并排除屈光参差及斜视的4～8岁儿童共1005例1910眼。采用Logistic多元回归分析年龄、性别、柱镜绝对值程度、球镜绝对值程度、散光类型对诊断子午线性弱视的影响，通过ROC曲线下面积（areaundertheROCcurve，AUC）分析进一步明确患者年龄因素对诊断子午线弱视的影响。结果：分别建立Logistic回归模型1（包括性别、柱镜绝对值程度、球镜绝对值程度、散光类型四个变量）和模型2（前四个变量再加上年龄）。两个模型的Logistic回归分析都提示柱镜绝对值程度是诊断子午线性弱视的影响因素，模型2的Logistic回归分析同时提示年龄是诊断子午线性弱视的影响因素。模型1的AUC为0．64，模型2的AUC为0．74，两者比较有统计学差异（P〈0．05），表明不同年龄儿童ROC的灵敏度和特异度存在一定差异。结论：运用Logistic回归和ROC曲线综合分析结果表明，在儿童子午线性弱视诊断中，需要充分考虑患者年龄因素对诊断的影响。
Bibliography:AIM： To assess the diagnostic value of age for meridional amblyopia by Logistic regression and receiver operating characteristic （ROC） curve. METHODS： A total of 1 005 children （1 910 eyes） with unknown ocular abnormalities other than with-the-rule astigmatism （ aged 4 - 8 years ） were recruited. Astigmatism 〉or = 1.00D and sphere 〈 or = 3.00D were present in one or both eyes. The difference of sphere between both eyes was less 1. 50D. The difference of astigmatism between both eyes was less 1. 00D. All astigmatism was calculated by the absolute value. By analyzing age, sex, astigmatism type, diopter of cylinder and diopter of sphere with Logistic regression, two mathematical models were established. Then the diagnostic efficacy of the model was assessed using the ROC curve. RESULTS： The model 1 included 4 parameters （sex, astigmatism type, diopter of cylinder and diopter of sphere）. The model 2 included 5 parameters （the 4parameters of the model 1 adding age）. Using Logistic regression, the diopter of cylind
ISSN:1672-5123
DOI:10.3980/j.issn.1672-5123.2013.07.40