A Multimodal Video-Based AI Biomarker for Aortic Stenosis Development and Progression

Abstract

Question: What is the association between a novel artificial intelligence video-based algorithm for aortic stenosis (AS) and disease progression?
Findings: In this multinational cohort study, a video-based biomarker of AS was reproducibly associated with a higher risk of developing new AS, faster progression of AS as shown by Doppler echocardiography, and future aortic valve replacement. The findings were replicated across 2 US-based echocardiography cohorts and in a cross-modality experiment using cardiac magnetic resonance videos in the UK Biobank.
Meaning: A unified video-based imaging biomarker may improve the risk stratification of AS development and progression across existing and widely performed modalities.