基于XGBoost的胸腰椎骨折内固定术后下肢深静脉血栓风险预测模型  

A XGBoost model for risk prediction of lower extremity deep vein thrombosis after internal fixation surgery for thoracolumbar fractures

作　　者：廖嘉佳 梁小娜 徐晓静 詹姜仙 LIAO Jiajia;LIANG Xiaona;XU Xiaojing;ZHAN Jiangxian(391Ward of Spine Surgery,the First Affiliated Hospital of Wenzhou Medical University,Wenzhou 325000,Zhejiang,China)

机构地区：[1]温州医科大学附属第一医院脊柱外科391病区,浙江温州325000

出　　处：《中国现代医生》2024年第33期47-51,116,共6页China Modern Doctor

摘　　要：目的基于极限梯度提升(extreme gradient boosting,XGBoost)构建胸腰椎骨折内固定术后发生下肢深静脉血栓(deep vein thrombosis,DVT)的预测模型。方法选取2019年1月至2022年12月于温州医科大学附属第一医院行胸腰椎骨折内固定术患者220例,分为训练集(154例)和测试集(66例)。训练集经过合成少数类过采样技术处理,基于XGBoost建立预测模型,在测试集上采用受试者操作特征曲线下面积、准确率、F1得分、敏感度和特异性指标比较性能,并基于SHAP值量化影响因素的贡献程度进行可解释性分析。结果XGBoost模型在多个指标上的表现优于逻辑回归、支持向量机和随机森林模型,其在原测试集上的曲线下面积为0.761,临床决策曲线表明其有一定的临床应用价值。结论基于年龄、体质量指数及术后白蛋白、D-二聚体、总蛋白、红细胞沉降率、凝血酶原时间建立的XGBoost模型可有效预测胸腰椎骨折内固定术后下肢DVT的发生,在临床实践中具有良好的应用前景。Objective To construct a predictive model for the occurrence of lower extremity deep vein thrombosis(DVT)after internal fixation surgery for thoracolumbar fractures by using extreme gradient boosting(XGBoost).Methods Data of 220 patients who underwent internal fixation surgery for thoracolumbar fractures in the First Affiliated Hospital of Wenzhou Medical University from January 2019 to December 2022 was collected.The dataset was divided into a training set(154 cases)and a testing set(66 cases).The training set was processed by using the synthetic minority over-sampling technique and the predictive model was build based on XGBoost.The performance was compared on the testing set by using area under receiver operating characteristic curve,accuracy,F1 score,sensitivity and specificity.The interpretability analysis base on SHAP was conducted to quantify the degree of contribution of influencing factors.Results The XGBoost model outperformed logistic regression,support vector machine and random forest models on multiple metrics,with an area under the curve of 0.761 on the original testing set.The decision curve indicated that the XGBoost model has clinical application value.Conclusion The XGBoost model based on factors such as age,body mass index,and postoperative albumin,D-dimer,total protein,erythrocyte sedimentation rate,prothrombin time can effectively predict the occurrence of lower extremity DVT after internal fixation surgery for thoracolumbar fractures,which has good potential for clinical application.

关 键 词：胸腰椎骨折 下肢深静脉血栓 合成少数类过采样技术 极限梯度提升 可解释性分析 

分 类 号：R619[医药卫生—外科学]