In a study that directly compared the Apple Watch Series 4 (AW4) with the KardiaBand (KB) for accuracy and sensitivity in detecting atrial fibrillation (AF), the automated algorithm used in the KB device outperformed that in AW4 on both measures. The findings of the so-called SMART WARS study were published in the journal JACC: Clinical Electrophysiology.
Both devices are capable of obtaining single lead electrocardiographic (ECG) recordings, a particularly unique opportunity for detecting paroxysmal arrhythmias, noted study authors, from the Department of Cardiology at Monash University, Melbourne, Australia.
Of key interest to the investigators in this study was to determine whether one device is better than the other at classifying AF correctly using an automated algorithm or using an interpretation augmented by a human.
The study was designed to obtain consecutive recordings from the AW4 and the KB concurrently with a 12-lead ECG. Investigators analyzed the automated diagnoses and blinded single-lead ECG tracing interpretations by 2 cardiologists as well as the effect of combined device and clinical interpretation.
In total, 125 patients were recruited from cardiology outpatient clinics between December 2019 and March 2020. Study inclusion required age ≥65 years and being able to tolerate lying supine for a 12-lead ECG. The rationale for the age cutoff, authors explain, is the increased prevalence of AF with age. Mean age among participants was 76±7 years and 62% were men.
The analysis found the accuracy of the automated rhythm assessment algorithm was greater with the KB device (74%) than with the AW4 device (65%). For the KB device, the authors report, the sensitivity and negative predictive value for AF were 89% and 97%, respectively; for the AW4 device, the values were 19% and 82%, respectively.
In models using hybrid automated and clinician interpretation, the overall accuracy of the KB device improved to 91% and the accuracy of the AW4 device increased to 87%.
“These findings suggest that although these devices’ tracings are of sufficient quality, automated diagnosis alone is not sufficient for making clinical decisions about atrial fibrillation diagnosis and management,” investigators wrote.
Marco V Perez, MD, a cardiac electrophysiologist and associate professor of medicine at Stanford Health Care , writing in an accompanying editorial, notes “The novelty of this study has to do with the machine-human hybrid approach. The idea that for complex problems, machine-assisted human interpretation will lead to optimal performance is growing.”
Reference: Ford C, Xie CX, Low A, et al. Comparison of 2 Smart Watch Algorithms for Detection of Atrial Fibrillation and the Benefit of Clinician Interpretation: SMART WARS Study. J Amm Coll Cardiol EP. 2022;8:782-791.