AI-Enhanced Digital Stethoscopes Help with Early Detection of Heart Disease

By Matt Phillion

A new research study has found that an AI-enabled digital stethoscope detected moderate to severe valvular heart disease with 92.3% sensitivity compared to 46.2% with traditional stethoscopes, more than doubling the detection rate in real-world primary care settings.

The study looked at 357 patients age 50 or over who were examined at three primary care clinics using both traditional and AI-enabled stethoscopes. The technology was used to listen for acoustic patterns associated with valve problems and flagged patients who need echocardiogram referrals.

Valvular heart disease affects more than half of adults over the ages of 65, but most cases go unreported through primary care. The condition can lead to heart failure, hospitalization, and can be fatal, but patients often display no symptoms until the disease is in advanced stages.

This new digital stethoscope isn’t intended to replace physicians, but rather enhances their clinical judgement and utilizes AI-trained pattern recognition human hearing may miss, explains Dr. Rose McDonough, the study’s senior author and director of medical affairs with Eko Health. Rather, it enables earlier detection that could reduce emergency hospitalization and overall costs—while also engaging patients more during exams.

“Seeing it in actual practice with patients, we saw substantial increases in detection for all types of murmur, mirroring our validation studies,” says McDonough.

While the stethoscope may have a different barrier to entry than many other emerging technologies given the familiarity of the tool, McDonough notes that what needs to happen it so make sure this technology can be merged seamlessly with current practices without adding time or work to clinicians and support staff.

“I’d argue that it’s easier because everyone knows the basic functionality of a stethoscope—the stethoscope is something you have on day one of medical school,” says McDonough. “It’s also familiar to the patient. We often see with AI that it causes more of a distance between the patient and the provider, but we’re bullish about our technology maintaining that human connection.”

It’s an interesting time in medicine, she explains, in that as more AI-enhanced tools become available, there is a tendency to discount the value of the physical exam.

“Studies show that cardiac auscultation skills among medical trainees have actually declined over time,” says McDonough. “And if you go to a provider now, because of everyone’s workload and other factors, you’re used to expecting a fast exam.”

The digital stethoscope requires you to take the time, to listen to a few breaths or heartbeats, and patients notice this. The tool also enables physicians to interact with patients in new ways, showing them what the technology picks up on from the exam.

“Overwhelmingly, patients feel comfortable with the stethoscope and they’re excited to have a more participatory experience. You can literally spin the screen around to show them, though you aren’t required to,” says McDonough. “You can see the results right away and show the patient the shape of their heartbeat and you can discuss the results you’re seeing.”

At this point it’s observational, McDonough notes, but there is something to the concept of the primary care doctor being able to have these direct, in-person interactions with the patient.

Clinician reactions

The deployments at this stage are adapted to each clinic. For many, it’s a medical assistant taking the recording rather than the physician, and then the provider can see the test results and re-listen if necessary.

“We don’t want to add to the workflow, and we want them to be able to maintain that connection. Observationally, they appreciate this opportunity for patient connection,” says McDonough.

The biggest impact so far, she notes, has been with primary care physicians. By the time a patient reaches a cardiologist, McDonough explains, you’re with a specialist who is more attuned and will often have imaging or other additional resources. They know the problems they’re looking for.

“In the primary care setting, it can be loud, there’s a lot going on, and this is a chance to have something that boosts their skills,” she says. “It’s not a replacement, it’s an augmentation that can alert them earlier on in the patient care journey and enables them to manage that care better.”

It offers a new entry point in that care journey and a way to get ahead because this disease can be so non-specific, McDonough explains.

“The patient might not notice they’re fatigued or short of breath,” she says. “You do this once a year during your annual wellness exam and you have a trajectory for the patient if a murmur is flagged.”

The device doesn’t yet grade on the severity of the murmur specifically, but it will pick up on mild as well as moderate or severe murmurs which can provide a chance to catch indications of the disease in its early stages.

“A positive exam can lead directly to a follow up test or cardiology referral, or it can flag a primary care provider to watch and monitor in their office,” she says.

Improving on the familiar

For this technology to be successful, McDonough explains, it needs to function as close to the traditional stethoscope as possible, which goes back to the issue of building it into the workflow in a non-intrusive way.

The other issue with getting it in the hands of as many clinicians as possible is reimbursement.

Eko has clearance for a new version of their algorithm that can get an even more refined picture of a patient’s heart health.

“It’s exciting both from a clinician trust perspective, but also for the patient really knowing what to do with that information,” McDonough says.

There’s also potential to look beyond cardiac use cases and explore other areas like lung or bowel exams where listening tools have applicability. Clinicians know how their stethoscope works, they know what it’s capable of, and this simply provides more firepower to provide even better care.

“The message we consistently try to provide is that this should make you a better clinician, to provide valuable insight to your patients, and improve patient outcomes,” says McDonough. “The vast majority of clinicians want this. They want to do the best they can and patients want the best care they can receive.”

It also flows into an issue every provider and healthcare organization is currently facing: There simply aren’t enough hours in the day and enough providers available to do everything they want to do.

“It’s something that can boost not only patient care, but clinician productivity,” says McDonough. “If you can get information faster to a provider, it tips them off to a potential diagnostic pathway to pursue sooner. This helps them not only get to the bottom of things faster but also helps streamline referrals.”

There’s so much overlap between heart and lung care, McDonough notes, and patients are often referred to both, so having a clearer sense at that first point of contact about what’s going on with the patient can thereby streamline referrals so patients are sent to the appropriate place.

“This feeds back into earlier detection of the disease. If you find it earlier, you’re less likely to end up at the hospital, for example,” she says.

It also plays a role in health equity. Consider patients who may not have access to the specialists they are being referred to or the ability to get to those specialists because of any number of social determinants of health.

“We want to empower providers. Primary care physicians in particular are overburdened, and so if we can give them more tools to help them be better clinicians, that’s huge,” says McDonough.

Matt Phillion is a freelance writer covering healthcare, cybersecurity, and more. He can be reached at matthew.phillion@gmail.com.