Columbia University researchers unveil EchoNext, an AI tool that analyzes ECG data to identify hidden structural heart disease, potentially transforming early diagnosis and treatment.
Artificial intelligence is revolutionizing heart disease screening, thanks to a new tool developed by researchers at Columbia University and NewYork-Presbyterian.
Structural heart disease, including conditions such as valve disease and congenital defects, often remain hidden until reaching advanced stages.
To address this, a research team led by Pierre Elias, an assistant professor of medicine and biomedical informatics at Columbia University Vagelos College of Physicians and Surgeons and medical director for artificial intelligence at NewYork-Presbyterian, developed EchoNext.
This AI-powered screening tool can analyze electrocardiogram (ECG) data to identify which patients require an echocardiogram, a more detailed ultrasound test.
In a landmark study published in the journal Nature, EchoNext showed superior accuracy in diagnosing structural heart disease compared to cardiologists.
“EchoNext basically uses the cheaper test to figure out who needs the more expensive ultrasound,” Elias said in a news release. “It detects diseases cardiologists can’t from an ECG. We think that ECG plus AI has the potential to create an entirely new screening paradigm.”
Transformative Potential
The typical ECG measures the heart’s electrical activity and is widely used to detect arrhythmias and other conditions. However, it traditionally cannot identify structural heart disease.
EchoNext changes that by leveraging AI to analyze ECG data, predicting when follow-up with an echocardiogram is necessary.
The tool was trained on a vast dataset of 1.2 million ECG-echocardiogram pairs from 230,000 patients.
In subsequent validation, it outperformed 13 cardiologists by accurately identifying 77% of structural heart issues, compared to the cardiologists’ 64%.
To test its real-world application, the researchers evaluated nearly 85,000 patients who underwent ECGs without prior echocardiograms. EchoNext flagged around 9% as high-risk for undiagnosed structural heart disease.
Follow-ups over a year revealed that nearly three-quarters of these high-risk individuals were diagnosed with such conditions, illustrating the tool’s efficacy.
Implications for Health Care
The introduction of EchoNext could significantly impact the early detection and treatment of structural heart disease.
With more than 400 million ECGs performed globally each year, this AI tool has the potential to convert every single one into an opportunity for diagnosis and timely intervention.
“You can’t treat the patient you don’t know about,” added Elias. “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.”
Future Directions
The researchers have made the deidentified dataset available to help other health systems enhance their heart disease screening processes. They are also currently running a clinical trial testing EchoNext across eight emergency departments.
Source: Columbia University