AI stethoscope flags heart failure and atrial fibrillation in seconds, study finds

An artificial intelligence–enabled stethoscope identified three heart conditions in as little as 15 seconds and substantially increased diagnosis rates for heart failure and atrial fibrillation, according to results from the TRICORDER study published in BMJ journals.

Researchers at Imperial College London and Imperial College Healthcare NHS Trust analyzed health records covering more than 1.5 million patients and examined 12,725 people with the new AI stethoscope technology. The study focused on patients presenting symptoms commonly associated with heart failure, including breathlessness, swelling and fatigue.

Patients evaluated with the AI stethoscope were about twice as likely to be diagnosed with heart failure as comparable patients who were not examined with the device. Those examined with the tool were roughly 3.5 times more likely to be diagnosed with atrial fibrillation, an abnormal heart rhythm linked to an increased risk of stroke. The study reported that the device could help clinicians identify three distinct cardiac conditions in each assessment, and that analysis took approximately 15 seconds per exam.

The British Heart Foundation, which partially funded the research, said the findings reflect the potential for novel diagnostic tools to detect serious cardiac disease at an earlier stage. The investigators drew on large-scale clinical data to compare diagnostic outcomes between patients assessed with the AI stethoscope and those who received standard care.

Traditional stethoscopes rely on clinician auscultation to detect abnormal heart sounds, while AI-enabled devices process acoustic and other sensor data to flag patterns that may indicate disease. In the TRICORDER study, the rapid output from the AI device was associated with higher rates of subsequent clinical diagnosis, suggesting it may serve as an adjunct to routine assessment in symptomatic patients.

Early diagnosis of heart failure can allow clinicians to begin therapies that slow disease progression and address symptoms; identifying atrial fibrillation enables treatment to reduce stroke risk. The study’s authors framed the results as evidence that an AI stethoscope can augment frontline detection of cardiovascular conditions, particularly among patients presenting with signs that warrant further cardiac evaluation.

The TRICORDER findings add to a growing body of research on artificial intelligence tools in clinical diagnostics. The study does not replace the need for confirmatory testing and clinical judgment, and researchers and funders have highlighted the importance of integrating new devices into care pathways with appropriate validation and oversight.