Machine learning-based screening of risk factors of early recurrence after surgery for concomitant exotropia and establishment of a Nomogram predic-tion model
10.13389/j.cnki.rao.2025.0021
- VernacularTitle:基于机器学习筛选共同性外斜视术后早期复发的风险因素及Nomogram预测模型的建立
- Author:
Jing XIE
1
;
Li PU
;
Zhengjing WANG
;
Hongfang HU
;
Liang FENG
;
Su ZHAO
Author Information
1. 550004 贵州省贵阳市,贵州医科大学附属医院眼科;550004 贵州省贵阳市,贵州医科大学;550005 贵州省贵阳市,贵阳爱尔眼科医院
- Publication Type:Journal Article
- Keywords:
concomitant exotropia;
early recurrence;
Nomogram prediction model;
machine learning
- From:
Recent Advances in Ophthalmology
2025;45(2):115-119
- CountryChina
- Language:Chinese
-
Abstract:
Objective To analyze the risk factors associated with early recurrence after surgery for concomitant exo-tropia and establish a Nomogram prediction model.Methods A retrospective analysis was conducted on 243 cases(486 eyes)of concomitant exotropia treated in the Ophthalmology Department of our hospital from October 2015 to October 2021.The patients were divided into a training set(n=170)and a validation set(n=73)at a ratio of 7∶3.The Lasso re-gression,Boruta algorithm,and random forest algorithm were used to screen risk variables related to postoperative recur-rence of concomitant exotropia.The Spearman correlation analysis and variance inflation factor(VIF)were used to assess collinearity among variables,and a Nomogram prediction model was established using multivariate Cox regression.The re-ceiver operating characteristic curve,calibration curve,and clinical decision curve of the model at 6 months,18 months,and 24 months after surgery were used to assess the efficacy of the model.Results Three machine learning methods in-cluding Lasso regression,Boruta algorithm,and random forest algorithm identified six significant variables that might con-tribute to early recurrence after strabismus surgery from 22 risk variables in both training and validation sets.No collineari-ty was found among the six variables(r＜0.6,VIF＜5).Multivariate Cox regression revealed that strabismus type(inter-mittent exotropia),preoperative strabismus angle,best-corrected visual acuity(BCVA)in the right eye,BCVA in both eyes,and surgical procedures(unilateral lateral rectus recession)were risk factors for early recurrence after surgery for concomitant exotropia.Meanwhile,a Nomogram prediction model was constructed based on these 6 factors.The receiver operating characteristic,calibration,and clinical decision curves indicated that the prediction model had good accuracy,consistency,and clinical applicability.Conclusion Nomogram prediction model can effectively predict the risk of early recurrence after surgery for concomitant exotropia,and provides a reference for ophthalmologists to intervene early in pa-tients.
