基于决策树法构建创伤性颅脑损伤术后硬脑膜下积液的风险预测模型  

Construction of a risk prediction model for post-traumatic subdural hygroma based on decision tree method

作　　者：张华 刘广明 刘国成 张姝红 陈大莉 蒲小龙 王志友 李倩 Hua Zhang;Guangming Liu;Guocheng Liu;Shuhong Zhang;Dali Chen;Xiaolong Pu;Zhiyou Wang;Qian Li(Department of Critical Medicine,Chengdu Pidu District People's Hospital,Chengdu 617300,China)

机构地区：[1]成都市郫都区人民医院重症医学科,617300

出　　处：《中华神经创伤外科电子杂志》2023年第1期19-25,共7页Chinese Journal Of Neurotraumatic Surgery：Electronic Edition

摘　　要：目的:探讨创伤性颅脑损伤(TBI)术后硬脑膜下积液(PTSH)的影响因素,并构建TBI术后PTSH的决策树风险预测模型。方法:选取成都市郫都区人民医院重症医学科自2019年6月至2022年6月收治的344例TBI患者作为研究对象,根据是否发生PTSH将其分为观察组和对照组,分析2组患者的临床资料,采用单因素和Logistic回归分析筛选TBI患者发生PTSH的影响因素,采用SPSS Modeler软件构建TBI术后PTSH的决策树预测模型,并分析决策树预测模型的诊断效能。结果:344例TBI患者行DC后发生PTSH者68例,发生率为19.77%;单因素和多因素Logistic回归分析结果显示,蛛网膜撕裂、入院GCS评分>8分、中线移位、大骨瓣、骨窗面积>150 cm2、骨瓣边缘距中线距离>2 cm是TBI患者行DC后发生PTSH的影响因素(P<0.05);决策树模型选出了蛛网膜撕裂、入院GCS评分>8分、中线移位、大骨瓣、骨窗面积>150 cm^(2)、骨瓣边缘距中线距离>2 cm等6个解释变量作为模型的节点,其中蛛网膜撕裂是最重要的预测因子。ROC曲线分析显示,决策树模型的曲线下面积(AUC)值为0.895,Logistic回归模型的AUC值为0.881,决策树模型的预测效能优于Logistic回归模型(Z=2.423,P=0.013)。结论:蛛网膜撕裂、入院GCS评分>8分、中线移位、大骨瓣、骨窗面积>150 cm^(2)、骨瓣边缘距中线距离>2 cm是TBI患者行DC后发生PTSH的影响因素,构建的决策树模型有助于筛查DC后发生PTSH高风险人群和指导临床制定科学的防治策略,具有较高的临床价值。Objective To investigate the influencing factors of post-traumatic subdural hygroma (PTSH) and to construct a decision tree risk prediction model for PTSH after traumatic brain injury (TBI) operation.MethodsFrom June 2019 to June 2022, 344 patients with TBI who were admitted to Intensive Care Medicine Department of Pidu District People's Hospital of Chengdu were selected as the study objects. According to whether PTSH occurred, they were divided into observation group and control group. The clinical data of the two groups were analyzed. The influencing factors of PTSH development in TBI patients were screened by univariate and Logistic regression analysis. The decision tree prediction model of PTSH after TBI was constructed by SPSS Modeler software. The diagnostic efficacy of the decision tree prediction model was analyzed.ResultsSixty-eight of 344 TBI patients developed PTSH after DC, with an incidence of 19.77%. The results of univariate and multivariate Logistic regression analysis showed that arachnoid tear, admission GCS score >8 points, midline shift, large bone flap, bone window area >150 cm2, distance from the edge of bone flap to the midline >2 cm were the influencing factors of PTSH in TBI patients after DC (P<0.05). Six explanatory variables were selected as nodes of the decision tree model, including arachnoid tear, admission GCS score >8 points, midline shift, large bone flap, bone window area >150 cm^(2), and distance from the edge of bone flap to the midline >2 cm, among which arachnoid tear is the most important predictor of the model. The ROC curve analysis showed that the area under curve (AUC) value of the decision tree model was 0.895, and the AUC value of the Logistic regression model was 0.881. The prediction efficiency of the decision tree model was better than that of the Logistic regression model (Z=2.423, P=0.013).ConclusionArachnoid tear, admission GCS score >8 points, midline shift, large bone flap, bone window area >150 cm^(2), and the distance from the edge of bone flap to the midlin

关 键 词：创伤后硬脑膜下积液 颅脑创伤 危险因素 决策树 预测模型 

分 类 号：R651.15[医药卫生—外科学]