Atrial Fibrillation Detection Using Smartwatch/Phone

Study Questions:

How does the accuracy of atrial fibrillation (AF) detection by smartwatch/phone technology compare with that of a standard electrocardiogram (ECG)?


One hundred patients (mean age, 68 years; 17 women) who were scheduled to undergo transthoracic cardioversion (CV) of AF were included. Prior to CV, both an ECG and a single-lead tracing (via the Kardia Band [KB]) were obtained in each patient. A 30-second tracing (equivalent to lead I) is obtained by the patient placing his/her thumb on an embedded electrode on the KB. (KB replaces the original band of the Apple Watch). The information is transmitted to an iphone via Bluetooth and analyzed by the Kardia application. An algorithm then categorizes the tracing as “possible AF,” “normal,” or “unclassified.” The tracings were downloaded from a secure server, printed, and then reviewed by blinded electrophysiologists.


After excluding eight patients who presented in sinus rhythm, 169 tracings and ECGs were available for review. Fifty-seven tracings were deemed “unclassified” by the algorithm, most often due to artifact and low-amplitude signals. As compared to the standard ECG, the sensitivity and specificity of the KB in diagnosing AF were 93% and 84%, respectively (Kappa coefficient, 0.77). The sensitivity and specificity associated with the physician’s analysis of the algorithm’s diagnosis were 99% and 83%, respectively (Kappa coefficient, 0.88).


The investigators concluded that the KB device and related technology, with physician oversight, can distinguish between AF and sinus rhythm.


For intermittent and infrequent arrhythmias, ambulatory monitoring is critical in establishing a rhythm diagnosis, which then guides patient management. Holter and event monitors are limited by finite duration of monitoring, cumbersome hardware, and dermatologic reactions. Implantable loop recorders are invasive and may also be limited by signal quality. The KB and the related KardiaMobile ECG (which is compatible with the Android platform) allow the patient to record a reasonable quality tracing during symptoms. The tracing can be reviewed with the physician in the office (via smartphone) and/or transmitted “instantaneously” to the physician’s office. Since the algorithm relies on rhythm irregularity, it is not known how it might fare with atrial flutter with a regular ventricular rate. Although the device/application seems relatively straightforward to use, the nontech-savvy elderly patient may find it daunting.

Keywords: ACC18, ACC Annual Scientific Session, Algorithms, Arrhythmias, Cardiac, Atrial Fibrillation, Atrial Flutter, Cell Phone, Electric Countershock, Electrocardiography, Electrodes, Monitoring, Ambulatory, Secondary Prevention

< Back to Listings