PO-626-01 PERFORMANCE AND ACCURACY OF A SMART WATCH SINGLE-LEAD ECG: A PILOT STUDY

Abstract

Prior studies have evaluated the accuracy of the Apple Watch single-lead ECG for the detection of atrial fibrillation and other arrhythmias. However, no study to date has analyzed the watch ECG morphology and interval characteristics compared to a standard 12 lead ECG, important in assessing patients on medications and for physician overreads. To analyze the characteristics of single-lead ECGs collected by an Apple Watch compared to a clinical 12 lead ECG. All Mayo Clinic ECG lab patients >18 years old were invited to participate in this study. Participants underwent a 30 second single-lead ECG recording with an Apple Watch Series 5 in a seated position within 5 minutes of 12 lead ECG (12 L). The collected ECGs were manually interpreted by physician staff. PR, QRS, QT intervals were manually calculated, and RR interval determined by computer. Seventy-four patients were included with 12 L ECG and Apple Watch ECG pairs. The cohort was 44.6% female (N=33) and 91.6% white (N=68) with an average age of 59.2. Six patients were in atrial fibrillation at the time of recording and six patients had pacemakers. The average differences between 12 L ECG and Apple Watch ECG intervals (in milliseconds) were: PR 23.3 ±18; QRS 18.9 ±16.3; QT 22 ±17.3; RR 79 ±70. Bland-Altman plots were created to visualize interval variation (Figure 1). Eleven patients were excluded from PR interval analysis as P waves were unable to be identified due to absence (i.e. atrial fibrillation) or other conditions (significant AV block, sinus bradycardia, low P amplitude in L I). Apple Watch ECG result interpretation was complicated by 1) aberrant inflection points before and after QRS complexes, 2) attenuation of P waves, 3) disappearance of pacemaker spikes and 4) significant artifact from a patient’s tremor (Figure 2). Apple Watch ECG processing can result in peri-QRS artifact, P-wave amplitude reduction, and pacemaker spike elimination. The Apple Watch single-lead ECG appeared most reliable in determining the QT interval and least reliable in determining the PR interval (Figure 1). The findings suggest that different processing or access to raw, unfiltered Apple Watch data may facilitate expert over-reads of these signals when applied for medical use.View Large Image Figure ViewerDownload Hi-res image Download (PPT)