Heart valve disorders are often referred to as a "silent epidemic" due to their widespread impact, affecting over half of individuals aged 65 and older. These conditions can remain unnoticed for years, as symptoms tend to be subtle or even absent. Researchers from the University of Cambridge have now created an AI-powered algorithm capable of detecting severe valve issues by analyzing heart sounds captured with a digital stethoscope.
The team recorded heart sounds from 1,800 elderly individuals and used echocardiography, the gold standard in diagnosing valve diseases, to verify the results. By training the algorithm with these audio recordings and the corresponding test results, the AI was able to recognize faint acoustic patterns that even seasoned doctors may miss.
The system demonstrated impressive accuracy, identifying 98% of severe cases of aortic stenosis—a potentially fatal condition that claims the lives of about 80% of untreated patients within two years. In total, the algorithm showed a sensitivity rate of 72% in detecting valve disease while suggesting unnecessary follow-up tests in only 18% of the cases. In comparison, doctors identifying the same recordings were only able to detect 62% of the cases and recommended unnecessary tests twice as often.
One of the key advantages of this AI system is its speed—requiring no more than 15 seconds of heart sound recording. This could lead to faster screenings, reduced costs, and a shorter wait time for patients needing an echocardiogram.
Informational material. 18+.
