Researchers have developed an AI system capable of identifying signs of heart disease from a routine ECG in under two seconds, drawing on patterns invisible to the human eye.
Standard electrocardiograms, which measure the heart’s electrical activity, have been used for a century to detect heart attacks and irregular rhythms. However, they can’t identify heart disease on their own. That typically requires an echocardiogram, an ultrasound scan patients often wait months to receive.
The new AI tool addresses that gap by analyzing ECG data to flag two of the most common types of heart disease: heart failure and heart valve disease. The findings were presented at the European Society of Cardiology’s annual meeting in Munich, one of the largest cardiology conferences in the world.
Because ECGs are performed roughly a billion times globally each year, researchers see major potential for widespread early detection. In a U.S.-based trial of 67,000 patients, the AI correctly identified up to 81% of heart failure cases and 90% of heart valve disease cases.
Dr. Sonya Babu-Narayan of the British Heart Foundation, which funded the trial, said the speed of the AI’s analysis is striking, comparing it to a “blink of an eye.” She noted that while the tool won’t catch every case, it could help prioritize patients most likely to have a heart condition — something that matters greatly, since earlier diagnosis and treatment improves survival outcomes.
Importantly, the AI isn’t meant to replace diagnostic testing. Instead, it would flag high-risk patients for expedited echocardiograms rather than leaving them on standard waiting lists that can stretch for months.
Professor Fu Siong Ng of Imperial College London explained that long wait times for heart ultrasounds make the tool especially valuable, since it could help identify and prioritize the patients most in need of urgent scans.
Beyond its primary use, Ng suggested the AI could also catch unsuspected cases, by screening all ECGs performed in a hospital for warning signs, even when a patient wasn’t originally being evaluated for heart disease.
Dr. Ahmed El-Medany, who led the analysis at Imperial College London, called the system a “superhuman AI” and said future work will focus on developing handheld ECG devices for healthcare workers.
Separately, researchers from the University of Tokyo and the Institute of Science Tokyo shared findings at the same conference showing that AI analysis of brief facial videos could help detect undiagnosed high blood pressure and type 2 diabetes — two conditions that frequently go unnoticed in millions of people worldwide.
