机器学习定位特发性室性心律失常起源的前瞻性研究
Prospective study of machine learning in locating the origin of idiopathic ventricular arrhythmia
傅国华 1郑建伟 2徐茵 1储慧民 1卓伟东 1王彬浩 1郁一波 1高昉 1丰明俊 1杜先锋 1孙一钧 1裘欣慧1
作者信息
- 1. 宁波大学附属第一医院心律失常诊疗中心,宁波 315000
- 2. 上海科心医学生物技术有限公司,上海20000
- 折叠
摘要
目的 评估高精度机器学习算法软件定位特发性室性心律失常(IVA)起源的准确性.方法 本研究为观察性描述性研究.入选2023年8月至10月在宁波大学附属第一医院心律失常诊疗中心行IVA导管射频消融术(RFCA)的患者,术前通过高精度机器学习算法软件预测IVA起源位置,与最终行RFCA成功的起源位置比较.结果 38例患者于术前进行起源位置预测,其中2例患者消融失败,最后共纳入首次成功行RFCA的36例IVA患者,其中男15例,年龄(53.1±9.6)岁.35例患者预测准确,1例IVA起源左心室顶部(Summit)内膜面的患者预测失败,软件预测准确率为97.22％(34/35).经过(3.1±0.3)个月随访,所有患者均达到长期消融成功标准.该机器学习算法软件预测IVA起源的受试者工作特征曲线下的面积、敏感度、特异度、F1评分、准确率分别为97％、97％、100％、96％、99％.结论 高精度机器学习算法软件定位IVA起源有效、精准.
Abstract
Objective To evaluate the accuracy of a high-precision machine learning algorithm software in locating the origin of idiopathic ventricular arrhythmia(IVA).Methods This observational study included patients who underwent IVA radiofrequency catheter ablation(RFCA)in Arrhythmia Center of The First Affiliated Hospital of Ningbo University from August to October 2023.Prior to the procedure,a high-precision machine learning algorithm software was used to predict the origin of IVA,and the predicted origin was compared with the actual origin determined during the successful RFCA procedure.Results A total of 38 patients had accepted prediction of the origin of IVA before procedure,of which 2 patients failed to ablate.Finally,a total of 36 patients with IVA who underwent successful RFCA procedures were included,with a mean age of(53.1±9.6)years,and 15 male patients(41.7％).Among them,35 patients had accurate predictions,with a software accuracy of 97.22％and 1 patient with IVA origin from summit had an inaccurate prediction.After a mean follow-up of(3.1±0.3)months,all the patients met the criteria for long-term ablation success.The area under the receiver operating characteristic curve,sensitivity,specificity,F1 score,and accuracy of the machine learning algorithm software in predicting the origin of ventricular arrhythmia were 97％,97％,100％,96％and 99％,respectively.Conclusion High-precision machine learning algorithm software effectively and accurately locates the origin of IVA.
关键词
人工智能/特发性室性心律失常/机器学习/射频导管消融/心电图Key words
Artificial intelligence/Idiopathic ventricular arrhythmia/Machine learning/Radiofrequency catheter ablation/Electrocardiogram引用本文复制引用
出版年
2024
NETL
NSTL
万方数据