The goal of AI is not to replace healthcare professionals but to enhance the tools they use, helping them rely on clinical judgment to detect what the human eye may miss. One of the most promising examples of this potential is the AI-powered stethoscope.

This is supported by the most recent evidence. Last February, a study published in European Heart Journal – Digital Health demonstrated that a digital stethoscope with AI more than doubles the sensitivity for detecting moderate-to-severe heart valve diseases in a real-world clinical setting compared with a conventional stethoscope.

To explore this issue further, El Médico Interactivo, part of the Medscape Professional Network, spoke with Jeffrey Lin, MD, board-certified cardiologist and chief medical officer of MDVIP, a national network of primary care doctors. The expert noted that key cardiovascular diseases — such as asymptomatic atrial fibrillation (AF), heart failure (HF), and valvular heart disease (VHD) — are still underdiagnosed. This is due, in part, to the fact that these conditions can be difficult to diagnose using traditional auscultation alone.

Early detection — typically when the patient is still seeing their primary care physician — is essential to preventing long-term problems, such as the progression of cardiac dysfunction and its complications. This is where AI-powered digital stethoscopes can provide significant value.

AI-Powered Digital Stethoscopes

“The most significant advancement lies in the ability of AI-powered stethoscopes to record multiple physiological signals, including not only heart sounds (phonocardiogram) but also ECG rhythm recordings,” explained Lin. This, combined with validated AI algorithms, can help diagnose conditions such as HF, AF, and VHD earlier and more accurately.

However, the specialist stressed a key principle: Clinical judgment remains indispensable when using any of these tools. While new technologies can help sharpen decision-making and support earlier diagnosis, they still need to be interpreted by a physician within the full context of the patient’s medical history.

Data, without the proper clinical context, can lead to confusion and unnecessary anxiety — a principle that is especially important in the case of false positives. Therefore, the stronger the doctor-patient relationship, the easier it will be to use this type of technology thoughtfully and clearly communicate the significance of a finding.

Implementation in Primary Care

In recent months, several studies have been published that point in this same direction and reinforce the potential of these devices in primary care settings. This is the case with the TRICORDER study, published in The Lancet earlier this year.

This is the first cluster-randomized controlled trial of clinical AI technology and the first large-scale trial conducted within a national primary care system. Its results demonstrated good diagnostic performance of AI-powered stethoscopes in detecting HF, AF, and VHD at the point of care.

“For primary care, this type of advancement is very promising. It helps drive a necessary paradigm shift, moving from a reactive system to a proactive model. As a cardiologist, I spent years observing the consequences of the accumulation of risks that were not detected in time, so any tool that helps us detect problems more quickly and act promptly is potentially very valuable.”

From Clinical Trial to Routine Practice

Although the evidence points to a potential benefit for physicians and patients, it remains to be seen how these tools will be implemented in clinical practice on an international scale.

“Adoption is still in its early stages. There is strong evidence supporting AI algorithms, but those same studies have highlighted difficulties in their implementation and adoption in practice,” reflected Lin.

To increase their use, the expert believes it is necessary to better understand the reasons that make it difficult for physicians to incorporate these tools and to facilitate their implementation by addressing issues such as integration with electronic health records, reimbursement, training, and the attitudes of healthcare professionals.

Ultimately, AI is poised to add a new layer of information to one of medicine’s most iconic tools. The challenge now will be to ensure that this innovation reaches the clinical setting in a way that is useful, integrated, and always at the service of clinical judgment.

Lin disclosed no relevant financial relationship.

This story was translated from El Médico Interactivo on Univadis.