AI derived ECG global longitudinal strain compared to echocardiographic measurements

Abstract

Abstract Left ventricular (LV) global longitudinal strain (LVGLS) is versatile; however, it is difficult to obtain. We evaluated the potential of an artificial intelligence (AI)-generated electrocardiography score for LVGLS estimation (ECG-GLS score) to diagnose LV systolic dysfunction and predict prognosis of patients with heart failure (HF). A convolutional neural network-based deep-learning algorithm was trained to estimate the echocardiography-derived GLS (LVGLS). ECG-GLS score performance was evaluated using data from an acute HF registry at another tertiary hospital (n = 1186). In the validation cohort, the ECG-GLS score could identify patients with impaired LVGLS (≤ 12%) (area under the receiver-operating characteristic curve [AUROC], 0.82; sensitivity, 85%; specificity, 59%). The performance of ECG-GLS in identifying patients with an LV ejection fraction (LVEF) < 40% (AUROC, 0.85) was comparable to that of LVGLS (AUROC, 0.83) (p = 0.08). Five-year outcomes (all-cause death; composite of all-cause death and hospitalization for HF) occurred significantly more frequently in patients with low ECG-GLS scores. Low ECG-GLS score was a significant risk factor for these outcomes after adjustment for other clinical risk factors and LVEF. The ECG-GLS score demonstrated a meaningful correlation with the LVGLS and is effective in risk stratification for long-term prognosis after acute HF, possibly acting as a practical alternative to the LVGLS.

Keywords

• Artificial intelligence
• Electrocardiography
• Global longitudinal strain
• Heart failure
• Prognosis