Automated Algorithm Using Smartwatch Technology Accurately Detects Atrial Fibrillation
The Kardia Band (KB) smartwatch automated algorithm for detection of atrial fibrillation (AFib), supported by physician review of the recordings, reliably differentiates AFib from sinus rhythm (SR), according to results of a single-center study presented by Joseph M. Bumgarner, MD, at ACC.18. The results were published simultaneously in the Journal of the American College of Cardiology.
The KB handheld cardiac rhythm recorder is a smartwatch accessory that records a rhythm strip equivalent to lead I for 30 seconds and has an algorithm for instantaneous diagnosis of AFib. The aim of this study was to examine whether the KB could differentiate SR from AFib compared with physician-interpreted 12-lead electrocardiogram (ECG) and KB recordings.
A total of 100 consecutive patients with AFib scheduled for cardioversion received a KB connected to a smartwatch paired via Bluetooth to a smartphone. The patients underwent pre-cardioversion ECG followed immediately by KB recording. If cardioversion was performed, post-cardioversion and KB recordings were obtained. The KB algorithm measures rhythm irregularity and P wave absence in real time to classify the rhythm strip as “possible AFib.” The automated KB interpretations were compared with electrophysiologist-interpreted KB rhythm strips and simultaneous ECGs.
Cardioversion was performed in 85 percent of the patients. A total of 169 simultaneous 12-lead ECG and KB recordings were obtained, of which 57 were unclassified by the KB algorithm. Of the classified KB strips, KB correctly diagnosed AFib with 93 percent sensitivity, 84 percent specificity and a K coefficient of 0.77 compared with electrophysiologist-interpreted 12-lead ECG. Electrophysiologist interpretation of KB recordings had 99 percent sensitivity, 83 percent specificity and a K coefficient of 0.83. Of the 57 KB unclassified recordings, the electrophysiologists correctly diagnosed AFib with 100 percent sensitivity, 80 percent specificity and a K coefficient of 0.74. Among interpretable KB recordings, the KB and electrophysiologist agreement was excellent (K coefficient 0.88).
While the results demonstrated the reliable differentiation of AFib from SR with the KB automated algorithm supported by physician review of the recordings, the authors note that the small sample size at their tertiary referral center had a known history of AFib with a burden sufficient to prompt electrical cardioversion. “The performance of the KB smartwatch algorithm may be more variable in a population with a lower AFib burden,” they wrote. Other factors that may affect the generalizability of the results are the low representation of women (17 percent), exclusion of patients with cardiac implantable electronic devices and lack of prior use the KB by participants. The Kardia Band has been approved by the U.S. Food and Drug Administration.
Keywords: ACC18, ACC Annual Scientific Session, Electric Countershock, Atrial Fibrillation, Sample Size, Tertiary Care Centers, United States Food and Drug Administration, Electrocardiography, Sensitivity and Specificity, Algorithms
< Back to Listings