构建心脏超声联合实验室指标的非瓣膜性心房颤动患者射频消融术后复发风险预测模型  

Constructing a Recurrence Risk Prediction Model for Non-Valvular Atrial Fibrillation Patients After Radiofrequency Ablation Using Echocardiography Combined with Laboratory Indicators

作　　者：邹文[1] 邓志生[2] 杨彦青 刘丽赟[4] 高泓 钟小亮[1] Zou Wen;Deng Zhisheng;Yang Yanqing;Liu Liyun;Gao Hong;Zhong Xiaoliang(Ultrasound Department,Nanchang First Hospital,Nanchang 330008,China;Geriatrics Department,Nanchang First Hospital,Nanchang 330008,China;Nursing Department,Nanchang First Hospital,Nanchang 330008,China;Cardiology Department,Nanchang First Hospital,Nanchang 330008,China)

机构地区：[1]南昌市第一医院超声科,330008 [2]南昌市第一医院老年医学科,330008 [3]南昌市第一医院护理部,330008 [4]南昌市第一医院心内科,南昌市330008

出　　处：《中国超声医学杂志》2024年第10期1111-1115,共5页Chinese Journal of Ultrasound in Medicine

基　　金：江西省卫生健康委科技计划项目(No.202311203)。

摘　　要：目的构建非瓣膜性心房颤动(NVAF)患者射频消融术(RFA)后复发风险模型并对其效能进行评价。方法回顾性分析首次接受RFA治疗的236例NVAF患者,通过单因素分析、LASSO回归及多因素向后逐步Logistic回归分析其独立影响因素并构建模型及绘制列线图,采用ROC曲线、校准曲线、Bootstrap法对模型进行评价及验证。结果心外膜脂肪组织厚度(EAT)、左心房前后径(LAD)、左心室射血分数(LVEF)、同型半胱氨酸(Hcy)是房颤复发的独立影响因素;模型ROC曲线AUC为0.889,灵敏度为84.9%,特异度为80.3%;Bootstrap法内部验证的AUC均值为0.897,Bootstrap法交叉验证的AUC为0.873,灵敏度为92.6%,特异度为78.3%,说明该模型具有较好的区分度;校准曲线平均绝对误差为0.026,Bootstrap法交叉验证的准确度为82.9%,Kappa值为0.742,说明该模型具有较好的准确度及一致性。结论基于EAT、LAD、LVEF及Hcy构建的NVAF患者RFA术后复发的预测模型列线图具有较好的区分度与准确度,对房颤复发有较高的预测价值。Objective To construct a recurrence risk model for patients with non-valvular atrial fibrillation(NVAF)after radiofrequency ablation(RFA)and evaluate its efficacy.Methods A retrospective analysis was conducted in 236 patients with NVAF who received RFA treatment for the first time.Univariate analysis,LASSO regression,and multiple factor stepwise Logistic regression were used to analyze the independent influencing factors,construct a model,and draw a column chart.ROC curve,calibration curve,and Bootstrap method were used to evaluate and validate the model.Results The results showed that epicardial adipose tissue thickness(EAT),left atrial diameter(LAD),left ventricular ejection fraction(LVEF),and homocysteine(Hcy)were independent factors affecting the recurrence of atrial fibrillation;The AUC of the ROC curve of the model was 0.889,with a sensitivity of 84.9%and specificity 80.3%;The average AUC of internal validation using Bootstrap method was 0.897,and the AUC of cross validation using Bootstrap method was 0.873.The sensitivity was 92.6%,and the specificity was 78.3%,indicating that the model had good discriminability;The mean absolute error of the calibration curve was 0.026,the accuracy of cross validation using Bootstrap method being 82.9%,and the Kappa value was 0.742,indicating that the model had good accuracy and consistency.Conclusions The column chart of the predictive model for postoperative recurrence in NVAF patients based on EAT,LAD,LVEF and Hcy has good discrimination ability and accuracy,and has high predictive value for atrial fibrillation recurrence.

关 键 词：预测模型 非瓣膜性心房颤动 射频消融术 心脏超声 复发 

分 类 号：R445.1[医药卫生—影像医学与核医学] R541.75[医药卫生—诊断学]