颅脑术后并发癫痫列线图预测模型的构建与应用  

Construction and application of a prediction model for postoperative epilepsy with a column chart after craniocerebral surgery

作　　者：杨燕妮 王雪华 Yang Yanni;Wang Xuehua(Department of Neurosurgery,the Second Affiliated Hospital of AFMU,Xi'an 710038,China)

机构地区：[1]空军军医大学第二附属医院神经外科,西安710038

出　　处：《脑与神经疾病杂志》2025年第2期67-71,共5页Journal of Brain and Nervous Diseases

基　　金：陕西省重点研发计划项目(2021SY-293)。

摘　　要：目的构建颅脑术后并发癫痫列线图预测模型并分析其应用价值。方法选择2021年12月至2023年12月于空军军医大学第二附属医院唐都医院收治的颅脑术后并发癫痫的患者86例为观察对象,纳为癫痫组(n=86),选择同期行颅脑术后未发生癫痫患者115例纳为非癫痫组(n=115),收集并比较两组患者临床资料及治疗方式情况,采用多因素Logistic回归分析颅脑术后并发癫痫的影响因素。采用R4.1.2软件包及MS程序包建立颅脑术后并发癫痫的风险预测列线图模型,采用受试者工作特征(ROC)曲线评估列线图模型预测颅脑术后并发癫痫的价值。结果癫痫患者术前格拉斯哥昏迷评分(GCS)评分13~15分、光反射(正常)占比明显低于非癫痫组,机械通气、脑疝、脑挫裂伤、颅内血肿占比明显高于非癫痫组(P<0.05)。癫痫患者修补材料自体占比明显高于非癫痫组(P<0.05)。多因素Logistic回归分析显示,GCS评分13~15分是颅脑术后并发癫痫的保护因素(P<0.05),机械通气、脑疝、颅内血肿是颅脑术后并发癫痫的危险因素(P<0.05)。根据多因素Logistic回归分析结果,通过R软件绘制颅脑术后并发癫痫的列线图模型,采用ROC曲线分析显示,颅脑术后并发癫痫的列线图模型预测颅脑术后并发癫痫的曲线下面积(AUC)为0.860(95%CI:0.778~0.941),敏感度为91.53%,特异度为70.00%。结论机械通气、脑疝、颅内血肿是颅脑术后并发癫痫的危险因素,GCS评分13~15分可降低癫痫发生风险,根据以上因素建立颅脑术后并发癫痫的列线图模型在预测颅脑术后并发癫痫方面具有一定价值,早期评价有助于指导治疗方案。Objective To construct a predictive model for postoperative epilepsy using a column chart and analyze its application value.Methods A retrospective analysis of clinical data was conducted.86 patients with postoperative epilepsy admitted to the Second Affiliated Hospital of AFMU from December 2021 to December 2023 were selected as the observation subjects,who were included in the epilepsy group(n=86).During the same period,115 patients without epilepsy who underwent craniocerebral surgery in our hospital were selected as the no epilepsy group(n=115).General information,clinical data,and treatment methods of the two groups of patients were collected and compared.Multivariate Logistic regression analysis was used to analyze the influencing factors of postoperative epilepsy.Establish a risk prediction column chart model for postoperative epilepsy using R4.1.2 software package and MS program package,and evaluate the value of the column chart model in predicting postoperative epilepsy using receiver operating characteristic(ROC)curve.Results The preoperative GCS score of epilepsy patients was 13-15 points,and the proportion of light reflex(normal)was significantly lower than that of the no epilepsy group.The proportion of mechanical ventilation,brain herniation,brain contusion,and intracranial hematoma was significantly higher than that of the no epilepsy group(P<0.05).The proportion of self repair materials in epilepsy patients was significantly higher than that in the no epilepsy group(P<0.05).Multivariate Logistic regression analysis showed that a GCS score of 13-15 was a protective factor for postoperative epilepsy(P<0.05),while mechanical ventilation,cerebral hernia,and intracranial hematoma were risk factors for postoperative epilepsy(P<0.05).According to the results of multiple Logistic regression analysis,a column chart model for postoperative epilepsy was drawn using R software.ROC curve analysis showed that the area under the curve(AUC)of the column chart model for predicting postoperative epilepsy was 0.860(

关 键 词：颅脑术后 癫痫 列线图 模型构建 临床应用 

分 类 号：R742.1[医药卫生—神经病学与精神病学]