基于临床和超声特点的二元Logistic 回归模型预测心脏黏液瘤血管栓塞事件  

Study on predicting vascular embolism events in cardiac myxoma with binarylogistic regression model based on clinical and ultrasound characteristics

作　　者：朱翔宇 叶红英 韩悦 王建美 ZHU Xiangyu;YE Hongying;HAN Yue;WANG Jianmei(Department of Ultrasound Medicine,Dezhou Hospital,Qilu Hospital,Shandong University,Dezhou 253075,China)

机构地区：[1]山东大学齐鲁医院德州医院超声医学科,山东德州253075

出　　处：《长春中医药大学学报》2024年第5期561-565,共5页Journal of Changchun University of Chinese Medicine

基　　金：山东省医药卫生科技发展计划项目(202104070512)。

摘　　要：目的基于临床和超声特点探究心脏黏液瘤(cardiac myxoma,CM)患者并发血管栓塞(vascular embolism,VE)的危险因素,并构建预测CM患者VE事件的二元Logistic回归模型。方法回顾性筛选2016年6月-2023年7月于山东大学齐鲁医院德州医院经病理证实为CM的患者120例,根据术前有无发生VE分为VE组和对照组,收集并对2组临床和超声心动图特点行单因素和多因素分析,建立Logistic回归模型,以采用受试者工作特征(ROC)曲线和Hosmer-Lemeshow检验评估模型的预测效能和拟合能力。结果单因素分析显示年龄、吸烟史、肿瘤形态和结构是CM患者VE发生的影响因素;多因素分析显示吸烟史、肿瘤形态不规则、肿瘤结构疏松是CM患者并发VE的独立危险因素。结论基于临床和超声特点建立的二元Logistic风险预测模型且具备较好的预测效能和拟合能力,可为心脏黏液瘤患者血管栓塞的预防提供一定的参考依据。Objective To explore the risk factors for vascular embolism(VE)in patients with cardiac myxoma(CM)based on clinical and ultrasound characteristics,and to construct a binary logistic regression model for predicting VE events in patients with CM.Methods The retrospective screening was conducted on 120 patients who were confirmed to have CM by pathology in Dezhou Hospital,Qilu Hospital,Shandong University from June 2016 to July 2023.Based on whether these patients experienced VE before surgery,they were divided into the VE group and the control group.Univariate and multivariate analyses on the clinical and echocardiographic characteristics of two groups were performed.A binary logistic regression model was established.The predictive efficacy and fitting ability of the model were assessed by using receiver operating characteristic(ROC)curve and Hosmer-Lemeshow test.Results The results of univariate analysis showed that age,smoking history,tumor morphology and structure were the factors affecting the occurrence of VE in CM patients;Multivariate analysis showed that smoking history,irregular tumor morphology,and lax tumor structure were independent risk factors for VE in CM patients.Conclusion A binary logistic risk prediction model established based on clinical and ultrasound characteristics has good predictive efficacy and fitting ability,which can provide the certain reference basis for the prevention of VE in the patients with CM.

关 键 词：超声心动图 心脏黏液瘤 血管栓塞 预测模型 

分 类 号：R743[医药卫生—神经病学与精神病学]