智能手表心电图应用于检测不同心室率患者心房颤动发作的诊断性能研究

Diagnostic performance of smart watch electrocardiograph in detecting atrial fibrillation episodes in patients with different ventricular rates

目的评估智能手表(Apple Watch)心电图应用在不同心室率患者中检测心房颤动(房颤)发作的诊断性能。方法入选2019年7月1日至2021年3月3日于上海交通大学医学院附属同仁医院就诊并完善12导联心电图检查且明确诊断为房颤或窦性心律的患者,同时完善智能手表心电图检查。根据12导联心电图的心室率将患者分为3组(A组,心室率<60次/min;B组,心室率60~100次/min;C组,心室率>100次/min)。电生理学家对12导联心电图及智能手表记录的单导联心电图分别判读。以电生理学家判读的12导联心电图结果为金标准,计算和比较智能手表单导联心电图的节律分类算法和电生理学家判读的智能手表心电图在不同心室率患者中检测房颤发作的灵敏度、特异度、Kappa值。结果共入组248例患者,其中男122例(49.2%,122/248),年龄(74.25±11.67)岁。12导联心电图诊断:129例为窦性心律,119例为房颤。与12导联心电图结果相比,智能手表单导联心电图的节律分类算法在A(41例)、B(163例)、C(44例)组患者中检测房颤发作的灵敏度分别为81.80%、63.83%、68.75%(Kappa值分别为0.98、0.70、0.55),特异度均为100%。而电生理学家判读的A、B、C组患者智能手表心电图在检测房颤发作的灵敏度(100.00%、88.20%、93.80%,Kappa值分别为1.00、0.88、0.89)比智能手表心电图节律分类算法明显提高,特异度均为100%。结论智能手表心电图应用检测房颤的节律分类算法仍需进一步改善。在低心室率患者中,使用智能手表检测房颤的临床效果最佳,为提高其诊断性能,智能手表心电图由电生理专家判读会更加可靠。

Objective To evaluate the diagnostic performance of the Apple Watch electrocardiograph(ECG)in detecting atrial fibrillation(AF)episodes in patients with different ventricular rates.Methods This study was conducted in Tongren Hospital,Shanghai Jiao Tong University School of Medicine between July 1,2019 and March 3,2021.Patients with AF or sinus rhythm were enrolled.Patients underwent a 12-lead ECG along with an Apple Watch ECG.The enrolled patients were divided into three groups(group A:ventricular rate below 60 bpm;group B:ventricular rate between 60 and 100 bpm;group C:ventricular rate above 100 bpm)according to the rhythm of the 12-lead ECG.All records of 12-lead ECG and Apple Watch ECG were interpreted by the electrophysiologist.The results of the 12-lead ECG interpreted by electrophysiologists were taken as the gold standard.The sensitivity,specificity,and K coefficient of the Apple Watch ECG rhythm algorithm to detect AF in patients with different ventricular rates were measured.In addition,the sensitivity,specificity and K coefficient of the Apple Watch ECG recordings interpreted by electrophysiologists for detecting AF in patients with different ventricular rates were also measured.Results A total of 248 patients were enrolled,122 were male,and the mean age was(74.25±11.67)years old.One hundred and twenty nine patients were diagnosed with sinus rhythm by the electrophysiologist interpretation according to 12-lead ECG,and 119 patients were diagnosed with AF.Compared with 12-lead ECG,the rhythm algorithm of Apple Watch ECG was 81.80%,63.83%and 68.75%sensitive for detecting AF and a specificity of 100%,100%and 100%in Group A,B,C(Group A including 41 patients,Group B including 163 patients,Group C including 44 patients;Kappa=0.98,0.70 and 0.55)respectively.The electrophysiologist interpretation of Apple Watch ECG recordings diagnosed AF with a sensitivity of 100.00%、88.20%、93.80%,and a specificity of 100%,100%and 100%in Group A,B,C(Kappa=1.00,0.88,0.89)respectively.The Apple Watch ECG recordings interpreted by the electrophysiologist had a higher sensitivity than the rhythm algorithm of Apple Watch ECG in detecting AF.Conclusion The rhythm algorithm of Apple Watch ECG application to detect AF still needs a further improvement.The Apple Watch ECG has a better diagnostic performance in detecting AF episodes in patients with low ventricular rate.In order to improve the diagnostic performance,Apple Watch ECG recordings would be more reliable when interpreted by the electrophysiologist.