AI-enhanced digital phonocardiography for TAVR procedural evaluation: Pilot study

Artificial intelligence (AI)–enhanced digital phonocardiography offers a noninvasive approach to evaluating valvular heart disease, yet changes in heart sounds following transcatheter aortic valve replacement (TAVR) remain underexplored. In this prospective, single-center pilot observational study, 28 patients with severe aortic stenosis undergoing elective transfemoral TAVR were enrolled. Pre- and immediate postprocedural heart sounds and electrocardiograms were recorded using AI-enhanced digital phonocardiography. Spectral area indices (SAIs), integrating murmur amplitude, duration, and frequency, were calculated for systole and diastole. Systolic-SAI was assessed in relation to stenosis severity, whereas diastolic-SAI was explored for post-TAVR paravalvular leak (PVL). A deep learning model incorporating heart sound and electrocardiogram waveforms was applied to classify stenosis severity pre- and post-TAVR. Echocardiographic and catheter-derived hemodynamic parameters served as references. After TAVR, systolic-SAI significantly decreased from 1.84 (1.31–2.37) to 1.04 (0.88–1.45) ( p < 0.001), paralleling hemodynamic improvement. AI-based binary classification showed favorable discrimination pre- and post-TAVR. Diastolic-SAI tended to be higher in patients with moderate versus trace/mild PVL, although statistical significance was not reached. AI-enhanced digital phonocardiography may enable quantitative assessment of aortic stenosis severity and immediate procedural outcomes in the catheterization laboratory and could complement echocardiography but requires external validation and larger multicenter studies. AI-enhanced digital phonocardiography may enable immediate functional assessment during TAVR. Spectral area index correlated with hemodynamic parameters and tracked procedural change. Digital phonocardiography may serve as a practical adjunct to echocardiography in minimalist TAVR pathways. • AI-enhanced phonocardiography enables immediate functional assessment during TAVR. • Spectral area index reflected hemodynamic parameters and procedural changes. • Digital phonocardiography can serve as a useful adjunct to echocardiography in TAVR.