Longitudinal Validation of a Deep Learning Index for Aortic Stenosis Progression

Background Aortic stenosis (AS) is a progressive disease requiring timely monitoring and intervention. While transthoracic echocardiography remains the diagnostic standard, deep learning–based approaches offer the potential for improved disease tracking. This study examined the longitudinal changes in a previously developed deep learning–derived index for AS continuum (DLi‐ASc) and assessed its prognostic association with progression to severe AS. Methods We retrospectively analyzed 2373 patients (7371 transthoracic echocardiographies) from 2 tertiary hospitals. DLi‐ASc (scaled 0–100), derived from parasternal long‐axis and short‐axis views, was tracked longitudinally. The median follow‐up duration was 42.8 (interquartile range, 22.2–75.7) months. Results DLi‐ASc increased in parallel with worsening AS stages ( P for trend<0.001) and showed strong correlations with aortic valve maximal velocity (Pearson correlation coefficient, 0.69; P <0.001) and mean pressure gradient (Pearson correlation coefficient, 0.66; P <0.001). Higher baseline DLi‐ASc was associated with a faster AS progression rate ( P for trend<0.001). Additionally, the annualized change in DLi‐ASc, estimated using linear mixed‐effect models, correlated strongly with the annualized progression of aortic valve maximal velocity (Pearson correlation coefficient, 0.71, P <0.001) and mean pressure gradient (Pearson correlation coefficient, =0.68; P <0.001). In Fine–Gray competing risk models, baseline DLi‐ASc was independently associated with progression to severe AS, even after adjustment for aortic valve maximal velocity or mean pressure gradient (hazard ratios per 10‐point increase, 2.38 and 2.80, respectively). Conclusions DLi‐ASc increased in parallel with AS progression and was independently associated with severe AS progression. These findings support its role as a noninvasive imaging‐based digital marker for longitudinal AS monitoring and risk stratification.