- A new study suggests that an artificial intelligence (AI) system can correctly identify 98% of patients with severe aortic stenosis and 94% with severe mitral regurgitation using short heart sound recordings from digital stethoscopes.
- Another study also suggests that AI stethoscopes are far more sensitive than traditional options and twice as efficient at detecting valvular heart disease in the clinic.
- The technology could offer a rapid, low-cost screening tool in primary care, helping identify patients who need echocardiography while reducing unnecessary referrals.
- By detecting subtle acoustic patterns, even in patients without obvious heart murmur, the AI could help diagnose valve disease earlier, when treatment is more effective and outcomes are better.
Heart valve disease occurs when one or more of the heart’s four valves
Research notes that valvular heart disease poses an
Diagnosing heart valve disease is often challenging, as people are typically asymptomatic until the condition advances, leading to delayed detection. Additionally, many symptoms are nonspecific or mistaken for usual aging.
Current diagnostic methods, such as traditional heart auscultation and
New research suggests that AI can consistently outperform clinicians and traditional stethoscopes for identifying heart valve disease, and deliver more reliable results, particularly for severe disease.
A study from the University of Cambridge, published in
In the study, researchers analyzed heart-sound recordings from 1,767 patients using an AI algorithm designed for digital stethoscopes. The participants were recruited across five National Health Service (NHS) Trusts. Each participant also underwent echocardiography, which served as a reference standard.
Rather than training the AI to identify heart murmurs, which clinicians traditionally listen for, the researchers trained it directly on echocardiogram results. This enabled the system to learn subtle acoustic patterns that may not be audible to the human ear, including in individuals without an obvious murmur.
The AI system accurately identified 98% of patients with severe aortic stenosis and 94% of those with severe mitral regurgitation.
When compared with 14 general practitioners (GPs), who assessed the same recordings, the AI system outperformed every clinician and produced consistent results. While individual GPs varied in how they balanced
The findings from the U.S. study suggest the AI stethoscope can more than double the sensitivity of routine clinical screening for moderate to severe valvular heart disease compared with a traditional acoustic stethoscope.
The prospective clinical investigation involved 357 adults ages 50 years and older attending routine primary care clinics. Participants were examined using both a traditional stethoscope and an AI-enabled digital stethoscope.
Heart sounds recorded by the AI device were analyzed using machine learning algorithms trained to recognize specific acoustic signatures associated with significant valvular dysfunction. The AI-enabled stethoscope had a sensitivity of 92.3%, while the traditional stethoscope had 46.2%.
This means that the AI device was able to correctly identify more than 9 out of 10 cases of moderate to severe disease, compared with roughly 4 to 5 out of 10 with the conventional tool.
Emileigh Lastowski, MS, Head of Clinical Research and Enablement at Eko Health, who was involved in the U.S. study, told Medical News Today:
“What stood out most was the magnitude of the difference in sensitivity between the AI-enabled digital stethoscope and the traditional stethoscope in a real-world clinical setting. We often expect controlled environments to show larger effects, so seeing this level of improvement during routine primary care exams was notable.”
— Emileigh Lastowski
“[AI]-enabled digital stethoscope exams can significantly improve the sensitivity of detecting moderate to severe valvular heart disease during routine clinical care, compared with traditional auscultation alone,” Lastowski said.
“This suggests that AI-assisted tools may help clinicians identify patients who would otherwise go undetected and refer them earlier for confirmatory testing,” she added.
Health experts often consider echocardiography the gold standard for diagnosing valve disease, such as aortic stenosis. However, it can be costly and time-consuming, making it unsuitable for widespread screening.
Although primary care physicians may listen to the heart using a stethoscope, it is not routine during short appointments. Additionally, it is a difficult skill, and clinicians can miss many cases.
As such, the AI-enhanced stethoscope can be a valuable tool for improving diagnostic accuracy, helping doctors decide which patients should be referred for further testing, and focusing resources on those who need them most.
The AI system was designed to minimize false positives. However, the US study observed a slight reduction in specificity. The authors suggest that this trade-off could be acceptable in screening settings where sensitivity is prioritised.
However, the authors stressed that AI is intended to support, not replace clinician judgement. Steven Steinhubl, MD, Chief Medical Officer at Eko Health, told MNT:
“What this study highlights is how clinically meaningful heart disease can go undetected during routine exams, not because clinicians are not skilled, but because valvular disease is often subtle and easy to miss in busy primary care settings.”
— Steven Steinhubl
“Early symptoms are frequently vague, and traditional auscultation depends on factors like experience, time spent on the physical exam, and even background noise,” Steinhubl continued.
“The findings suggest that our AI-enabled digital stethoscopes can provide meaningful decision support at the point of care, helping clinicians surface risk earlier and make more informed referral decisions during everyday visits,” he said.
Although the AI-supported stethoscope shows promise, both studies highlight further research across more diverse clinical environments and populations is necessary. Additionally, while the AI stethoscope can detect severe forms of valvular heart disease, detecting moderate forms remains more challenging.
However, experts believe AI-based tools could help address increasing pressures on healthcare systems caused by aging populations. Utilizing simple, scalable screening tools such as the AI-enhanced stethoscope could help by identifying individuals before irreversible damage occurs.
“As healthcare systems face aging populations and ongoing workforce constraints, tools that strengthen routine exams without adding complexity may play an important role in improving detection while preserving clinical workflow,” Steinhubl said.