AI-enabled ECG tool detects aortic stenosis years before intervention, study finds

AccurKardia’s artificial intelligence–enabled electrocardiogram technology can detect aortic stenosis years before patients require valve replacement and help predict outcomes after treatment, according to a new study published in European Heart Journal – Digital Health.

The study, led by Matthew Segar, an electrophysiology fellow at the Texas Heart Institute, evaluated the company’s AK-AVS algorithm across community-based populations and patients who later underwent transcatheter aortic valve replacement (TAVR) at Baylor St. Luke’s Medical Center. Researchers found the AI-enabled ECG model was able to identify signs of aortic stenosis as much as 4.5 years before TAVR intervention, suggesting the technology could enable earlier detection and more accessible monitoring.

Aortic stenosis is among the most common and serious valvular heart diseases in older adults and can lead to heart failure, hospitalization and death if not diagnosed and treated early.

The analysis also showed that patients who screened positive for aortic stenosis using the AI-ECG model but did not yet show disease on echocardiography had a 4.4-fold increased risk of future hospitalization for the condition over a median follow-up of 6.2 years. Researchers said the findings indicate the algorithm may identify early electrical changes in the heart before structural abnormalities become visible with conventional imaging.

In addition, AI-ECG trajectory patterns independently predicted increased one-year mortality risk following TAVR, identifying risks not captured by widely used clinical risk scores such as the Society of Thoracic Surgeons and EuroSCORE models.

“This study demonstrates that AK-AVS could not only enable earlier detection of aortic stenosis, but it may also be a useful tool in surveillance and predicting outcomes,” said David Shavelle, chief of cardiology for the MemorialCare Health System.

Segar said the technology “has the potential to transform how clinicians screen, monitor, and risk-stratify patients,” helping physicians intervene earlier and improve outcomes.

Because ECGs are inexpensive and widely available, researchers said AI-enhanced ECG analysis could expand screening and risk assessment across large patient populations. AccurKardia said the next step will be real-world pilot programs designed to identify undiagnosed aortic stenosis patients and guide treatment decisions.

Advances accelerate detection and treatment of valvular heart disease

Technological advances in cardiovascular diagnostics and treatment are rapidly reshaping how clinicians detect and manage aortic stenosis and other valvular heart diseases. Historically, diagnosis has relied primarily on echocardiography, often performed after patients develop symptoms such as shortness of breath or chest pain. Increasingly, however, earlier detection strategies are moving toward routine screening tools augmented by artificial intelligence.

AI-based ECG analysis is emerging as a particularly promising approach because ECG testing is inexpensive, widely performed and already embedded in routine clinical workflows. Algorithms trained on large datasets are being developed to identify subtle electrical patterns associated not only with aortic stenosis but also with conditions such as reduced ejection fraction, atrial fibrillation risk and structural heart disease. These tools aim to flag high-risk patients for follow-up imaging, potentially shifting care from symptom-based diagnosis to proactive surveillance.

At the same time, treatment advances — especially the widespread adoption of transcatheter aortic valve replacement — have dramatically expanded options for patients who were previously considered too high-risk for open-heart surgery. TAVR procedures have evolved from a therapy primarily for frail or inoperable patients to one increasingly used across broader risk categories, supported by improvements in device design, procedural techniques and imaging guidance.

Researchers are also working to integrate predictive analytics into peri-procedural planning and post-procedure monitoring. Risk-prediction models that combine clinical data, imaging and AI-derived biomarkers may help physicians better determine the optimal timing of valve replacement and identify patients who require closer follow-up after intervention.

Together, advances in AI-driven diagnostics, minimally invasive valve therapies and predictive risk modeling are contributing to a shift toward earlier detection and more personalized management of valvular heart disease, with the goal of reducing hospitalizations, improving survival and maintaining quality of life as populations age.