Human-monitored ECG outperforms AI in detection of significant arrhythmias

Carnation Ambulatory Monitor

Long-term continuous electrocardiogtraphy (LT-ECG), coupled with oversight by a physician, significantly outperforms algorithm-dependent mobile cardiac telemetry (MCT) in the detection of significant arrhythmias according to the findings of a study presented during a late-breaking trial session at the 2021 Heart Rhythm Society annual meeting (HRS 2021, 28–31 July, Boston, USA and virtual).

The study, presented by author Mark Willcox (Alaska Heart & Vascular Institute, Alaska, USA) sought to compare the “real-world” diagnostic accuracy of two US Food and Drug Administration (FDA) approve cardiac monitors—the BioGuardian algorithm-based MCT (Preventice) and the Carnation ambulatory monitor (BardyDx)—by using both devices simultaneously in the same patient population.

This study was conducted in an outpatient arrhythmia clinic and enrolled 50 sequential patients who simultaneously wore both the MCT and a the long-term continuous ECG monitor. Of the 50 enrolled patients, four failed to wear both monitors simultaneously and were excluded from the study. Patients wore both monitors simultaneously for at least some time period (range: 1.2‒14.8 days). Each study and its associated report were reviewed by two electrophysiologists and were categorised based on whether significant clinical arrythmias were identified and correctly diagnosed.

Willcox reported that during the simultaneous recording, significant arrhythmias were diagnosed by MCT in 11/46 patients (24%) compared to 23/46 patients by the long-term continuous ECG monitoring, a 209% increase (p=0.018).

In addition, in 12 of the 46 patients (26%), a significant arrhythmia finding was missed by MCT but was captured by long-term ECG. In two patients, AV node re-entrant tachycardia, captured by long-term ECG, was missed by MCT. In three patients, second degree AV block was unreported by MCT but captured by long-term continuous ECG monitoring. In 7/46 (15%), ventricular tachycardia (VT) was reported by MCT, compared to 13/46 (28%) patients by long-term continuous ECG monitoring.

Atrial fibrillation (AF) was documented by both types of monitors in two patients, however, long-term continuous ECG monitoring captured four additional AF episodes missed by MCT.

“Reports received from cardiac monitors greatly impact and influence clinical decisions, so accurate readings are crucial. Because every cardiac monitor employs different reporting methods to process recorded rhythms, it can be challenging for the physicians receiving and interpreting the final report. Our study set out to better understand the accuracy between two common options” said Willcox. “We found technology, paired with key human oversight and input, proved to be the most accurate in detecting critical arrhythmias and, ultimately, helping to improve patient outcomes.”

The authors hope the results of their study will lead to further conversation into the approval process for monitoring devices and how to ensure reporting accuracy among available monitors.


Please enter your comment!
Please enter your name here