AI Screening Tool Revolutionizes Heart Disease Detection with Low-Cost ECG Data

New Delhi: Researchers at Columbia University have developed an artificial intelligence (AI) tool called EchoNext that uses low-cost electrocardiogram (ECG) data to detect structural heart diseases with greater accuracy than cardiologists. The findings were published in the journal Nature.

Structural heart disease, which includes valve disease, congenital heart defects, and other conditions affecting heart function, often goes undiagnosed due to the lack of affordable and routine screening methods. EchoNext aims to address this issue by analyzing ordinary ECG data to identify patients who require further testing with an echocardiogram, a non-invasive ultrasound used to diagnose structural heart problems.

EchoNext was trained on a dataset of over 1.2 million paired ECG-echocardiogram records from 230,000 patients. In validation studies conducted across four hospital systems, including NewYork-Presbyterian campuses, the AI model demonstrated high accuracy in detecting a range of heart conditions such as cardiomyopathy-related heart failure, valve disease, pulmonary hypertension, and severe heart thickening.

In a direct comparison involving 13 cardiologists reviewing 3,200 ECGs, EchoNext correctly identified 77 percent of structural heart problems, outperforming cardiologists who achieved a 64 percent accuracy rate based on ECG data alone.

Pierre Elias, Assistant Professor of Medicine and biomedical informatics at Columbia University Vagelos College of Physicians and Surgeons, explained the significance of the tool: “We have colonoscopies, we have mammograms, but we have no equivalents for most forms of heart disease.”

EchoNext is designed to use the cheaper ECG test to determine which patients need the more expensive ultrasound, Elias added: “It detects diseases cardiologists can't detect from an ECG. We think that ECG plus AI has the potential to create an entirely new screening paradigm.”

Elias also highlighted the potential impact of the technology, stating, “Using our technology, we may be able to turn the estimated 400 million ECGs that will be performed worldwide this year into 400 million chances to screen for structural heart disease and potentially deliver life-saving treatment at the most opportune time.”

The development of EchoNext offers a promising step towards improving early detection and management of structural heart diseases, potentially benefiting millions globally by making screening more accessible and accurate.