Essayagh, Benjamin A

Abstract Aims Prognosis in tricuspid regurgitation (TR) is shaped by complex clinical and haemodynamic interactions, complicating risk stratification. We aimed to use explainable artificial intelligence (EAI) to model these relationships and identify key predictors of outcomes. Methods and results We analysed long-term all-cause mortality in a large prospective registry of patients with native TR across the full severity spectrum. Patients underwent detailed clinical evaluation and quantitative Doppler-echocardiography. Supervised EAI models were applied to identify and rank predictors of mortality under medical therapy, capturing non-linear associations and interactions among baseline characteristics. Among 9761 patients [mean age 73 ± 15 years; 56.9% women; effective regurgitant orifice area (EROA) 40 (30–56) mm2; regurgitant volume 39 (29–53) mL/beat; 42% severe TR], 2676 deaths occurred during 3.8 ± 4.6 years of follow-up. EAI identified and ranked 20 prognostic determinants of survival. Haemodynamic measures, led by elevated pulmonary-artery systolic and right-atrial pressures, were strongest, followed by age and comorbidity. TR-related variables ranked next, led by right ventricular dysfunction and EROA, which outperformed integrative TR grade and TRIScore. Left-ventricular parameters were less influential. Non-linear effects, including those involving ejection fraction, and interactions such as higher EROA-related risk with elevated pulmonary pressures or lower TRIScore (both P < 0.03) were observed. Proportional hazards models confirmed EROA as the strongest TR severity marker (P < 0.0001), with excess mortality beginning at 40 mm2 and increasing further at 60 mm2. Conclusion EAI ranked mortality predictors in TR, highlighting patient factors, haemodynamics, TR-specific measures, and left-ventricular function. Among TR-specific measures, EROA and TRIScore were most prognostic. Recognizing their impact can refine risk stratification, guide intervention timing, and inform long-term management.

Abstract Aims Early detection of aortic stenosis (AS) is becoming increasingly important with a better outcome after aortic valve replacement in asymptomatic severe AS patients and a poor outcome in moderate AS. We aimed to develop artificial intelligence-enabled electrocardiogram (AI-ECG) using a convolutional neural network to identify patients with moderate to severe AS. Methods and results Between 1989 and 2019, 258 607 adults [mean age 63 ± 16.3 years; women 122 790 (48%)] with an echocardiography and an ECG performed within 180 days were identified from the Mayo Clinic database. Moderate to severe AS by echocardiography was present in 9723 (3.7%) patients. Artificial intelligence training was performed in 129 788 (50%), validation in 25 893 (10%), and testing in 102 926 (40%) randomly selected subjects. In the test group, the AI-ECG labelled 3833 (3.7%) patients as positive with the area under the curve (AUC) of 0.85. The sensitivity, specificity, and accuracy were 78%, 74%, and 74%, respectively. The sensitivity increased and the specificity decreased as age increased. Women had lower sensitivity but higher specificity compared with men at any age groups. The model performance increased when age and sex were added to the model (AUC 0.87), which further increased to 0.90 in patients without hypertension. Patients with false-positive AI-ECGs had twice the risk for developing moderate or severe AS in 15 years compared with true negative AI-ECGs (hazard ratio 2.18, 95% confidence interval 1.90–2.50). Conclusion An AI-ECG can identify patients with moderate or severe AS and may serve as a powerful screening tool for AS in the community.