For the past decade, Eko Health has been advancing the use of artificial intelligence to help providers detect the early signs of disease, particularly heart and lung conditions, during routine checkups.
The digital health company gave a centuries-old device—the stethoscope—an AI upgrade to revamp it as a connected medical device that functions as an early disease detection platform.
Last April, the Food and Drug Administration cleared Eko's AI algorithm that captures cases of low cardiac ejection fraction, a key indicator of developing heart failure. That follows previous FDA nods for its AI-powered algorithms that detect structural heart murmurs and atrial fibrillation.
The company continues to push forward to use AI and smart stethoscopes to bridge the gaps in cardiac care.
Working with Brown University Health System’s Cardiovascular Institute, Eko Health developed an algorithm for the detection of pulmonary hypertension (PH), which is characterized by elevated pressure in the blood vessels connecting the heart and lungs.
If left untreated, pulmonary hypertension can lead to heart failure, early disability, or death. Despite being a serious condition, PH is often underdiagnosed due to the limited availability of effective detection tools.
An analysis of the algorithm demonstrated its ability to analyze heart sounds recorded with a digital stethoscope for identifying elevated pulmonary artery systolic pressures, a key indicator of PH, according to a peer-reviewed study published in the Journal of the American Heart Association (JAHA).
Leveraging 6,000 heart-sound recordings paired with echocardiographic pressure estimates to train the AI model, the algorithm demonstrated significant performance, with an average area under the receiver operating characteristic (AUROC) curve of 0.79, a sensitivity of 71%, and a specificity of 73%, according to the company.
RELATED: Eko Health captures $41M to expand AI stethoscope business worldwide
"This innovative approach demonstrates how combining digital stethoscopes with advanced AI can lead to a low-cost, non-invasive, point-of-care screening tool for the early detection of pulmonary hypertension," Gaurav Choudhary, M.D., Lead Principal Investigator, and Ruth and Paul Levinger Professor of Cardiology at the Alpert Medical School of Brown University, said in a statement. "Our findings represent a significant advancement in clinical practice that can ultimately enhance patient care."
Ongoing data collection from over 1,200 patients continues to refine the model’s accuracy and clinical utility, with the goal of further improving detection capabilities for broader clinical use, Eko Health said.
The study underscores the potential of Eko Health's non-invasive, rapid detection tool to aid clinical decision-making in primary care and other settings where costly or invasive diagnostic methods are less accessible. It opens the door to use Eko Health's AI stethoscopes to potentially screen for patients outside of clinical settings in the future, the company's executives said.
Additionally, the algorithm demonstrated its ability to pinpoint specific, clinically relevant segments of heart sound recordings, offering a transparent and explainable AI approach that aligns with physicians’ diagnostic workflows.
"The company’s goal is to develop pioneering AI solutions that address significant gaps in healthcare delivery," said Steve Steinhubl, M.D., Chairman of Eko’s Scientific Advisory Board, in a statement. "Early detection and intervention are essential in addressing cardiovascular diseases, and Eko is dedicated to providing accessible and scalable technologies that empower healthcare providers while improving patient care globally."
Modernizing early disease detection during routine checkups
While excitement over artificial intelligence in healthcare has reached a fever pitch, Eko Health has been using AI since the company's inception.
Connor Landgraf co-founded Eko Health in 2013 with his University of California, Berkeley classmates Jason Bellet and Tyler Crouch after seeing the limitations of traditional stethoscopes while studying biomedical engineering. Landgraf observed the limitations of the traditional stethoscope while interviewing cardiologists on their clinical pain points. He saw the potential to upgrade the analog stethoscope to develop a smart device.
The company's first product was an attachment for traditional stethoscopes with a compatible smartphone application.
"The very first deck we built almost 10 years ago, we called it the 'Shazam for heartbeats.' That was the pitch from day one," Landgraf told Fierce Healthcare, referring to the music identification app. "I think we were probably early. AI wasn't even really as a buzzword at the time, but we knew that algorithms and Shazam-like technologies could work and could be very accurate, and we were going to be building them for heart and lung sounds using this really unique data type."
RELATED: Eko secures FDA clearance for stethoscope AI to catch low ejection fraction
Eko Health started building algorithms for detect heart murmurs and atrial fibrillation. Its murmur detection algorithm was clinically validated in a Massachusetts General Hospital study and found to double identification rates of structural heart murmurs versus conventional practice in primary care. The low EF detection algorithm, developed with Mayo Clinic, was shown in an Imperial College London study to significantly enhance the identification of heart failure with reduced ejection fraction in GP clinics.
The company then expanded its capabilities into early detection of heart failure.
"This is particularly novel because it's a signal that would typically be undetectable to a physician with a traditional stethoscope. This signal is so subtle that a provider might not even be able to identify it themselves. This is a first-of-its-kind, 15-second non-invasive early detection test to identify patients who have a weak heart pump," he said. "We're continuing to expand the roadmap. We're knocking over the pins or the algorithms that are relatively easy and straightforward and then we're starting to move into the ones that are able to identify signs that are so subtle that the physicians can't even spot them themselves."
The company's aim is to revolutionize diagnostics using AI to more quickly detect heart issues when patients have their first point of contact with a clinician.
"These AI tools are most impactful when they're deployed in primary care or frontline care settings where access to other tests is much more limited. It's unlikely that you're going to be able to access an echocardiogram in your primary care physician's office, and every time you add a referral or an order or test that has to have follow-up, you miss some fraction of patients. Some fraction of patients doesn't go to see the cardiologists when they've been referred," Landgraf said.
"Every time we add those hurdles in to accessing care, the number of patients who complete it, the number of patients who follow through, goes down. So, building the tools that can be used right there on the spot by the primary care physician, by the frontline healthcare professional, and given them more insight into that decision-making process. Put an orange flag for this patient. Say, 'You should spend more time with this patient. This patient should be much more cautiously managed. You should talk to this patient about heart disease risk.' We think that could be pretty impactful and changing the course of the disease progression for the patients."
Heart disease remains the leading cause of death in the United States, accounting for about 1 in every 5 deaths, according to the CDC. Early detection plays a critical role in improving patient outcomes.
Eko Health now offers a growing portfolio of digital stethoscopes, patient and provider software and AI-powered analysis. It claims its FDA-cleared platform is used by over 500,000 healthcare professionals worldwide.
The digital health company has raised more than $165 million in funding, including a $41 million series D round in June. The company is backed by investors ARTIS Ventures, DigiTx Partners, Double Point Ventures, EDBI, Highland Capital Partners, LG Technology Ventures, Mayo Clinic, Morningside Technology Ventures Limited, NTTVC, Questa Capital and others.
In November, the company was granted a Category III CPT code from the American Medical Association (AMA) for its Sensora platform, which includes FDA-cleared AI algorithms that identify signs of structural heart murmurs, low ejection fraction (Low EF), and arrhythmias, including AFib. That milestone puts Eko Health's AI on a path to reimbursement.
"The AMA's creation of Category III CPT code for Eko's AI disease detection algorithms is a major step in increasing access to early heart disease detection," Landgraf said.