B-PO01-098 REAL-LIFE PERFORMANCE, LONG-TERM ROBUSTNESS, AND ABSENCE OF RACE BIAS IN THE ARTIFICIAL INTELLIGENCE ENHANCED ELECTROCARDIOGRAM FOR THE DETECTION OF LEFT VENTRICULAR SYSTOLIC DYSFUNCTION

We previously reported on the application of artificial intelligence (AI) to a standard ECG (AI-ECG) to detect left ventricular systolic dysfunction (LVSD). This tool remains promising for a rapid, inexpensive, point of care screening strategy for patients with previously undiagnosed LVSD. Some AI models require ongoing re-training, and have racial bias. We assessed the AI-ECG for detecting EF≤40% with respect to race and ethnicity, time, and...