HeartBeam has announced new study data demonstrating no significant differences in the detection of atrial fibrillation (AF), atrial flutter and sinus rhythm when the company’s deep learning algorithms were applied to the HeartBeam system, or to standard 12-lead electrocardiograms (ECGs).
The new data were presented by Joshua Lampert (Mount Sinai Hospital, New York, USA) during HRX Live 2025 (4–6 September, Atlanta, USA).
In the study, ECG recordings were taken with both the HeartBeam system—which captures the heart’s electrical signals from three distinct directions—and a standard 12-lead ECG in 201 consecutive patients. Deep learning algorithms previously developed from more than 10,000 standard 12-lead ECGs were applied to both sets of recordings and compared against diagnoses made by an expert panel of three electrophysiologists.
The study found no significant differences in multiple accuracy measures between the HeartBeam group and the standard 12-lead ECG group across diagnoses of 131 sinus rhythms, 57 AFs, and 13 atrial flutters. In addition, high accuracy rates for detecting AF, atrial flutter and sinus rhythm were observed in both groups (94.5% HeartBeam vs 95.5% standard 12-lead ECG).
“This study represents an exciting step forward in making advanced cardiac monitoring more user-friendly and widespread,” said Rob Eno, chief executive officer (CEO) of HeartBeam. “The comparable performance of deep learning algorithms applied to HeartBeam’s credit card-sized device with 3D, non-coplanar signals and traditional 12-lead ECG systems for detecting common arrhythmias like atrial fibrillation and flutter opens new avenues for patient care, particularly in settings where a full standard 12-lead ECG might be impractical.”
HeartBeam plans to use data from its deep learning algorithms, including from this study, to support future US Food and Drug Administration (FDA) submissions to enhance forthcoming product offerings.
