AI-Enabled Direct-from-ECG Ejection Fraction (EF) Severity Assessment Using COR ECG Wearable Monitor

Summary

This prospective, multicenter, cluster-randomized controlled study aims to evaluate the accuracy of an investigational artificial intelligence (AI) Software as a Medical Device (SaMD) designed to compute ejection fraction (EF) severity categories based on the American Society of Echocardiography's (ASE) 4-category scale. The software analyzes continuous ECG waveform data acquired by the FDA-cleared Peerbridge COR® ECG Wearable Monitor, an ambulatory patch device designed for use during daily activities. The AI software assists clinicians in cardiac evaluations by estimating EF severity, which reflects how well the heart pumps blood. In this study, EF severity determination will be made using 5-minute ECG recordings collected during a 15-minute resting period with participants seated upright. The results will be compared to EF severity obtained from an FDA-cleared, non-contrast transthoracic echocardiogram (TTE) predicate device. This comparison aims to validate the accuracy of the AI software.

Detailed Description

Objective This prospective study benchmarks the accuracy of CorEFS AI software in estimating ejection fraction (EF) severity categories using continuous ECG waveforms from the FDA-cleared Peerbridge Cor® ECG device, calibrated to the American Society of Echocardiography (ASE) scale. Background Heart failure (HF) remains a significant public health issue, particularly in older adults (75+), with high morbidity and mortality rates. Half of HF cases involve reduced EF (HFrEF), a condition associated with a 75% five-year mortality rate. Despite advancements in HF management, accessible, low-cost EF monitoring is lacking. Echocardiography (Echo) is the gold standard for EF measurement but is limited in ambulatory and home settings. Continuous ECG wearables like the Peerbridge Cor® offer a promising alternative, providing high diagnostic yield, low wear burden, and real-time EF estimation. Previous studies (References 1-11) demonstrate the potential of AI-enabled ECG analysis in EF prediction, with accuracies up to 91.4% and AUCs of 0.94 in estimating EF severity. Successful demonstration of the proposed endpoints to clinically acceptable statistical thresholds will provide a new and alternative capability for EF severity assessments compared to ultrasound, MRI, and other imaging modalities where access is limited. Hypothesis Specific ECG changes may identify left ventricular dysfunction (LVSD) and predict EF severity, enabling low-burden, cost-effective EF monitoring in high-risk populations. Study Design Participant Enrollment and Setup Participants will receive the Peerbridge Cor® wearable, with data collection occurring through: In-clinic setup: Study staff apply and initiate device use. Patient Home Setup (PHS): Telehealth guidance for independent device application (20% of participants). Subprotocols A: 30 minutes of Cor® ECG recording; 15 minutes analyzed. B: Up to 7 days of Cor® device use with periodic 15-minute sitting sessions. EF Reference Standard EF severity will be determined via FDA-cleared transthoracic echocardiography (TTE), using the Simpson's Bi-Plane Method. Data Collection Peerbridge Cor® ECG Data: 30 minutes recorded; 15 minutes analyzed in 5-minute segments. Echo Study: Conducted before or during Cor® recording. 12-Lead ECG: Simultaneous recording with the Cor® device. Participants log sessions using the Cor® device's Event button. De-identified medical histories will support subgroup analyses. Endpoints Agreement between Cor® ECG-derived EF severity and Echo results will be assessed across ASE-defined categories (Normal, Mild, Moderate, Severe). Positive predictive value (PPV) adjusted for prevalence will be calculated. This streamlined protocol validates CorEFS software for reliable, cost-effective EF monitoring and clinical decision support.

Eligibility Criteria

Interventions