目的：基于XGboost算法构建青光眼患者术后发生干眼的风险预测模型。

方法：回顾性选取我院2022年7月至2025年6月收治的青光眼术后患者为研究对象，根据术后是否发生干眼分为发生干眼组和未发生干眼组。收集患者临床资料，采用单因素分析及多因素分析患者发生干眼的影响因素。另按7：3的比例随机分为训练集和验证集，以影响因素为特征变量构建XGboost风险预测模型，使用SHapley值加性解释(SHAP)条形图和蜂群图可视化XGboost模型。采用受试者工作特征曲线(ROC)评估模型的预测效能。

结果：本研究纳入青光眼术后患者300例300眼，未发生干眼组204例204眼(男104例，女100例)，发生干眼组96例96眼(男55例，女41例)，干眼发生率为32.0%。单因素分析显示，两组患者年龄、合并糖尿病、睑脂黏度分级、泪膜破裂时间、睑板腺功能障碍、手术时间比较均有差异(均P<0.05)。多因素分析显示，年龄、合并糖尿病、睑脂黏度分级、泪膜破裂时间、睑板腺功能障碍、手术时间为青光眼患者术后发生干眼的影响因素(均P<0.01)。构建XGboost模型，结果显示，发生干眼的影响因素按重要性排序由高至低依次为泪膜破裂时间、合并糖尿病、手术时间、年龄、睑板腺功能障碍、睑脂黏度分级。ROC分析显示，XGboost模型训练集AUC为0.84(95%CI：0.78-0.90)，验证集AUC为0.83(95%CI：0.74-0.92)(均P<0.05)。

结论：基于XGboost算法构建的青光眼患者术后发生干眼的风险预测模型预测性能较好，临床或可据此识别高风险术后干眼患者，给予针对性干预措施以预防。

METHODS:A retrospective analysis was performed on glaucoma patients who received surgical treatment at the hospital from July 2022 to June 2025. All patients were divided into a dry eye group and a non-dry eye group according to the occurrence of postoperative dry eye disease. Clinical data of the patients were collected, and univariate and multivariate logistic regression analyses were employed to screen out the risk factors for postoperative dry eye. The patients were randomly allocated into a training set and a validation set at a ratio of 7:3. An XGboost risk prediction model was built with the risk factors as feature variables, and the SHapley Additive exPlanations(SHAP)bar plot and beeswarm plot were used for visual interpretation of the model. The predictive efficacy of the model was evaluated via receiver operating characteristic(ROC)curve analysis.

RESULTS:The study included 300 glaucoma patients(300 eyes). The non-dry eye group comprised 204 patients(204 eyes, 104 males and 100 females), and the dry eye group comprised 96 patients(96 eyes, 55 males and 41 females). The incidence rate of postoperative dry eye was 32.0%. Univariate analysis revealed statistically significant differences between the two groups in terms of age, comorbid diabetes, meibum viscosity grade, tear film breakup time, meibomian gland dysfunction, and operative time(all P<0.05). Multivariate logistic regression analysis showed that all the above factors were risk factors for postoperative dry eye(all P<0.01). The XGBoost model showed that these risk factors were ranked in descending order of predictive importance as: tear film breakup time, comorbid diabetes, operative time, age, meibomian gland dysfunction, and meibum viscosity grade. ROC curve analysis demonstrated that the area under the curve(AUC)of the XGboost model was 0.84(95%CI: 0.78-0.90)for the training set and 0.83(95%CI: 0.74-0.92)for the validation set, with both values showing statistical significance(both P<0.05).

CONCLUSION:The XGboost algorithm-based risk prediction model for postoperative dry eye in glaucoma patients exhibits favorable predictive performance. It can be clinically applied to identify patients at high risk of developing postoperative dry eye, thereby facilitating targeted interventions for preventive purposes.