A new artificial intelligence system can detect serious heart valve disease with greater accuracy than general practitioners using a standard stethoscope, offering the potential for earlier diagnosis and treatment of a condition often dubbed a “silent epidemic.” Researchers at the University of Cambridge, collaborating with clinicians from five NHS Trusts, developed the AI, which analyzes heart sounds to identify valve disease, a condition affecting over half of those aged 65 and older.
The study, published in npj Cardiovascular Health, analyzed heart sounds from 1,767 patients, comparing the AI’s performance against echocardiograms – the current gold standard for diagnosis. The AI correctly identified 98% of patients with severe aortic stenosis, a narrowing of the aortic valve, and 94% of those with severe mitral regurgitation, where the mitral valve doesn’t close properly. These are the most common forms of valve disease requiring surgical intervention.
“Valve disease is a silent epidemic,” said Professor Anurag Agarwal from Cambridge’s Department of Engineering, who led the research. “An estimated 300,000 people in the UK have severe aortic stenosis alone, and around a third don’t recognize it. By the time symptoms appear, outcomes can be worse than for many cancers.”
Currently, diagnosing valve disease relies heavily on echocardiography, a process that can be expensive and involve lengthy waiting times within the National Health Service. Traditional stethoscope-based screening, although readily available, is often limited by the skill of the practitioner and the constraints of short GP appointments. “Cardiac auscultation is a demanding skill, and it’s used less and less in busy GP surgeries,” Agarwal explained. “That’s a big part of why so many cases of valve disease are being missed.”
The Cambridge team’s AI differs from previous attempts to use artificial intelligence in this field by directly analyzing echocardiogram results rather than focusing on detecting heart murmurs, the traditional diagnostic marker. This approach allowed the system to identify subtle acoustic patterns that might be missed by human listeners, even in cases where no murmur is present. A study published in Open Heart in 2023, led by Geoff Strange, also highlighted the potential of AI-driven analysis of echocardiographic measurements to identify severe aortic stenosis phenotypes associated with high mortality.
In testing, the AI consistently outperformed 14 GPs who evaluated the same heart sound recordings. While GPs varied in their diagnostic approaches – some prioritizing sensitivity, others specificity – the AI delivered reliable results every time, demonstrating particular accuracy in identifying severe cases. The system was designed to minimize false alarms, a crucial consideration given the potential strain on echocardiography services.
The technology is intended to serve as a screening tool, assisting doctors in identifying patients who require further investigation, rather than replacing them entirely. “If you can rule out people who definitely don’t have significant disease, you can focus resources on those who need them most,” Agarwal said. The test requires only a few seconds of heart sound recording and can be administered by staff with minimal training.
Researchers acknowledge that further trials are needed in real-world GP settings with a diverse patient population before the device can be widely implemented. They also note that detecting moderate forms of valve disease remains a challenge. A recent article in Nature detailed the development and validation of a recurrent neural network achieving an AUROC of 0.83, demonstrating sensitivity for severe aortic stenosis (98%) and severe mitral regurgitation (94%).
Professor Rick Steeds, from University Hospitals Birmingham and a co-author of the study, emphasized the importance of timely intervention. “Valve disease is treatable. We can repair or replace damaged valves and give people many more years of healthy life,” he said. “But timing is everything. Simple, scalable screening tools like this could make a real difference by finding patients before irreversible damage occurs.”
The research was supported by the National Institute for Health Research, the British Heart Foundation, and the Medical Research Council.
