FDA Approves AI Stethoscope for Earlier Heart Failure Detection

Stethoscope on smartphone | Image credit: AndSus – stock.adobe.com

The FDA today approved Eko Health's stethoscope with embedded artificial intelligence (AI) to detect heart failure (HF) earlier during routine checkups.

According to a news release, Eko Health's Low Ejection Fraction (Low EF) detection AI works in 15 seconds via an Eko stethoscope, marking “both a significant medical innovation and a new era in the detection of cardiovascular disease.”¹

There are more than 6 million individuals living with HF in the US, with half of these individuals having HF with reduced EF (HFrEF), which affects the heart's ability to pump blood efficiently. Traditional methods like echocardiography—often inaccessible in primary care provider offices due to cost and training requirements—contribute to cases going undiagnosed until symptoms necessitate specialized care or emergency visits, leading to poorer outcomes and increased health care expenses. Built in collaboration with Mayo Clinic, Eko Health’s Low EF AI aims to change this by integrating quick and accessible low EF detection into routine stethoscope exams at the frontline of health care.

“The ability to identify a hidden, potentially life-threatening heart condition using a tool that primary care and subspecialist clinicians are familiar with—the stethoscope—can help us prevent hospitalizations and adverse events,” Paul Friedman, MD, chair of the Department of Cardiovascular Medicine at Mayo Clinic, said in the release. “Importantly, since a stethoscope is small and portable, this technology can be used in urban and remote locations, and hopefully help address care in underserved areas.”

The Eko Low EF AI is designed to assist clinicians in detecting individuals with a left ventricular EF (LVEF) of 40% or less. It analyzes electrocardiogram (ECG) and heart sound recordings from patients using signal processing and machine learning algorithms, leveraging data from FDA-cleared Eko Health stethoscopes that are stored in the Eko Cloud. The AI tool also provides machine learning-based notifications indicating the likelihood of LVEF of 40% or less, prompting further referral or diagnostic follow-up. It’s important to note that the newly approved AI serves as a screening tool and does not replace diagnostic assessments by health care professionals, offering an additional layer of detection for previously undiagnosed LV dysfunction during physical examinations.

Research presented at the American Heart Association’s Scientific Sessions in November 2023 included nearly 1200 Nigerian women during pregnancy or postpartum and demonstrated that peripartum cardiomyopathy was identified twice as frequently when ECG testing was conducted with an Eko digital stethoscope employing an AI algorithm compared with standard clinical ECG along with routine obstetric care.² The study also identified pregnancy-related cardiomyopathy in 4% of women during pregnancy and postpartum who were screened using the AI-enabled digital stethoscope, whereas the detection rate was 1.8% in the control group, indicating that about half of the cases of cardiomyopathy may go unnoticed with standard care.

Another study published in The Lancet Digital Health revealed that the AI tool could identify individuals with reduced LVEF (≤40%) using single-lead ECG inputs.³ Among 1050 patients undergoing transthoracic echocardiogram, various models achieved area under the receiver operating characteristic curves (AUROCs) ranging from 0.85 to 0.91, indicating promising diagnostic accuracy. These findings suggested that by integrating ECG recording and AI, the stethoscope examination—a routine component of clinical practice—can serve as a screening tool for detecting reduced LVEF at the point of care.

“Given the frequent clinical encounters of undiagnosed patients before index hospital admission for heart failure, the stethoscope examination has the potential to be a point-of-care screening opportunity, and through further AI algorithms, to become a tool for comprehensive detection of cardiovascular disease,” the Lancet study authors wrote.³

The AI tool also underwent rigorous training and validation processes.⁴ It was trained using a comprehensive dataset comprising more than 100,000 pairs of ECGs and echocardiograms from patients. Clinical validation was conducted in a multisite prospective study involving 3456 patients, where it achieved an AUROC of 0.835 for detecting LVEF less than 40%, demonstrating a robust ability to distinguish between low and normal EF with a sensitivity of 74.7% and specificity of 77.5%.

According to the Eko Health news release, the addition of the Low EF AI to Eko's Sensora Cardiac Early Detection Platform enhances its capabilities, complementing existing FDA-cleared algorithms designed to detect atrial fibrillation and structural heart murmurs, commonly associated with valvular heart disease.¹ When low EF is identified during a primary care examination using Sensora, it facilitates prompt referral to the cardiology department for comprehensive diagnostic assessments and treatment evaluation, potentially accelerating access to life-prolonging therapies.

References

1. FDA clears first AI to aid heart failure detection during routine check-ups. News release. Eko Health. April 2, 2024. Accessed April 2, 2024. https://www.ekohealth.com/blogs/newsroom/fda-clears-low-ejection-fraction-ai
2. AI technology improved detection of heart disease during and after pregnancy. News release. American Heart Association. November 13, 2023. Accessed April 2, 2024. https://newsroom.heart.org/news/ai-technology-improved-detection-of-heart-disease-during-and-after-pregnancy
3. Bachtiger P, Petri CF, Scott FE, et al. Point-of-care screening for heart failure with reduced ejection fraction using artificial intelligence during ECG-enabled stethoscope examination in London, UK: a prospective, observational, multicentre study. Lancet Digit Health. 2022;4(2):e117-e125. doi:10.1016/S2589-7500(21)00256-9
4. 510(k) premarket notification: K233409. FDA. Accessed April 2, 2024. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm?ID=K233409