Prospective Study of Deep Learning Algorithms to Improve Echocardiogram Analysis for Heart Valve Disease at Columbia University

What this Trial Is About

Heart disease is the leading cause of death in the United States, and echocardiography echo is the most common method doctors use to examine the heart. Valve diseases such as aortic stenosis and mitral regurgitation are common in older adults but often go undiagnosed. This research aims to test whether artificial intelligence AI can help cardiologists analyze echocardiograms more quickly and consistently by providing real-time AI analysis during echo reading. Researchers at Columbia University have developed deep learning algorithms that interpret echocardiograms by classifying views, identifying structures, measuring heart dimensions, and diagnosing valve disease. This study will continue to develop these AI models using patient data and evaluate an automated system that delivers AI analyses to cardiologists while they review echocardiograms. The goal is to improve the accuracy and efficiency of echo readings. Participants are attending cardiologists who read transthoracic echocardiograms at Columbia University facilities and have consented to participate. The study involves reviewing echocardiograms with AI support and comparing physician and AI assessments, especially for cases of moderate to severe valve leakage or narrowing. Outcomes measured include how often AI changes clinical decisions or leads to referrals for specialists or further testing over an 18-month period. Expert panel reviews and follow-up assessments will monitor the impact of AI on clinical management.

DELINEATE-Prospective