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

The FDA today approved the first artificial intelligence (AI) tool to detect heart failure (HF) early during routine exams.

Eko Health’s low ejection fraction (Low EF) detection AI operates in 15 seconds through the Eko stethoscope, ushering in “important medical innovation and a new era in cardiovascular disease detection,” according to a news release. is showing.¹

More than 6 million people in the United States have heart failure, and half of them have heart failure with reduced EF (HFrEF), which affects the heart’s ability to pump blood efficiently. Traditional methods such as echocardiography are often not available in primary care provider offices due to cost and training requirements, resulting in cases going undiagnosed until symptoms require specialized care or emergency visits. This can lead to poor prognosis and increased medical costs. Eko Health’s Low EF AI, built in partnership with Mayo Clinic, aims to change this by integrating rapid and accessible Low EF detection into routine stethoscope examinations on the front lines of healthcare. We are aiming for

“The ability to identify potentially life-threatening hidden cardiac conditions using a stethoscope, a familiar tool for primary care physicians and subspecialist clinicians, will help prevent hospitalizations and adverse events.” said Dr. Paul Friedman, chair of the department. The Mayo Clinic Department of Cardiology said in a release. “Importantly, because the stethoscope is small and portable, this technology can be used in urban and remote areas, and is expected to help address care in underserved areas. ”

The Eko Low EF AI is designed to help clinicians detect individuals with left ventricular EF (LVEF) below 40%. Leveraging data from her FDA-cleared Eko Health stethoscope stored in Eko Cloud, it uses signal processing and machine learning algorithms to analyze electrocardiogram (ECG) and heart sound recordings from patients. The AI ​​tool also provides machine learning-based notifications that indicate that the likelihood of LVEF is below 40%, prompting further referrals and diagnostic follow-up. The newly approved AI serves as a screening tool and does not replace a diagnostic evaluation by a medical professional, but rather provides an additional layer to detect previously undiagnosed left ventricular dysfunction during a physical exam. It is important to note that

The study, presented at the November 2023 American Heart Association Scientific Sessions, included nearly 1,200 pregnant or postpartum Nigerian women who underwent ECG examinations with the Eko digital stethoscope, which employs AI algorithms. When performed, it was demonstrated that peripartum cardiomyopathy was identified twice as often compared to standard clinical ECG. In parallel with regular obstetric care.² The study also identified pregnancy-related cardiomyopathy in 4% of pregnant and postpartum women screened using an AI-enabled digital stethoscope, compared to a detection rate of 1.8% in the control group. , indicating that approximately half of cardiomyopathy cases may be pregnancy-related cardiomyopathy. It goes unnoticed with standard care.

Another study published in lancet digital health showed that this AI tool was able to identify people with reduced LVEF (≤40%) using single-lead ECG input.³ Among 1,050 patients who underwent transthoracic echocardiography, various models achieved area under the receiver operating characteristic curve (AUROC) ranging from 0.85 to 0.91, demonstrating promising diagnostic accuracy. These findings suggested that by integrating ECG recordings and AI, stethoscope testing, a routine component of clinical practice, can serve as a screening tool to detect decreased LVEF at the point of care.

“Given the frequent clinical encounters of undiagnosed patients before admission with heart failure, stethoscope examination represents a point-of-care screening opportunity and, through further AI algorithms, a comprehensive detection tool for cardiovascular disease. ”, lancet the study authors wrote.³

The AI ​​tool also underwent a rigorous training and validation process.^Four It was trained using a comprehensive dataset consisting of over 100,000 pairs of ECGs and echocardiograms from patients. Clinical validation was performed in a multicenter prospective study of 3,456 patients, achieving an AUROC of 0.835 for detection of LVEF <40% and distinguishing between low and normal EF with sensitivity of 74.7% and specificity of 77.5%. demonstrated a robust ability to .

The addition of Low EF AI enhances Eko’s Sensora Cardiac Early Detection Platform to detect atrial fibrillation and structural heart murmurs commonly associated with valvular heart disease, according to a news release from Eko Health. Complements existing FDA-approved algorithms designed for¹ If low EF is identified during primary care testing using Sensora, prompt referral to cardiology for comprehensive diagnostic and treatment evaluation can be facilitated, facilitating access to life-sustaining treatments. There is a gender.

References

1. The FDA has approved the first AI to help detect heart failure during routine exams. news release. eco health. April 2, 2024. Accessed April 2, 2024. https://www.ekohealth.com/blogs/newsroom/fda-clears-low-ejection-fraction-ai
2. AI technology has improved the 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 due to reduced ejection fraction using artificial intelligence during electrocardiogram-enabled stethoscope examination in London, UK: A prospective observational multicenter study. lancet digit health. 2022;4(2):e117-e125. doi:10.1016/S2589-7500(21)00256-9
4. 510(k) Premarket Notification: K233409. F.D.A. Accessed April 2, 2024. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm?ID=K233409