The U.S. FDA has given the greenlight to Eko Devices Inc.’s electrocardiogram (ECG)-based algorithm to aid in detecting patients with heart failure during the COVID-19 pandemic. The artificial intelligence (AI)-powered algorithm, which provides a quick way to screen for low ejection fraction, won FDA breakthrough status in December of 2019.
The emergency use authorization is limited to the use of the ECG algorithm by a health care provider for purposes of assessing left ventricular ejection factor (LVEF) as an aid in screening for potential cardiac complications associated with COVID-19 or underlying heart conditions that may impact patient management. LVEF is a sign that the heart’s pumping mechanism is not doing its job, and not enough blood is leaving the heart each time it contracts.
“The need for cardiac screening for patients at risk of heart disease has never been more important,” Jason Bellet, Eko co-founder and chief operating officer, told BioWorld. The EUA is specifically aimed at adults with known or suspected COVID-19 infections.
High risk of death
Heart failure affects about 5.7 million people in the U.S. and 26 million worldwide. People with a history of heart disease are at greater risk of the coronavirus, which can cause heart attack, arrhythmia and a weakened heart muscle. The American College of Cardiology has estimated the death rate in COVID-19 patients with preexisting cardiovascular disease at 10.5%.
“Given the danger COVID-19 poses to patients with a weak heart pump, it’s important that we rapidly identify these individuals early and monitor them closely,” said Paul Friedman, head of cardiovascular medicine at Mayo Clinic, which collaborated on the development of the algorithm. “By embedding the heart failure screening AI into a quick, widely available and safe test using existing medical devices, we can detect heart failure early and start appropriate treatments.”
Eko’s ECG algorithm analyzes 15 seconds of ECG data collected with the Eko Duo digital stethoscope and, within seconds, displays a binary prediction of the patient’s likelihood of LVEF less than or equal to 40% on a smartphone or other device.
“It is groundbreaking that an AI-based algorithm can identify signs of ventricular dysfunction in a 12-lead ECG, taking it beyond the ECG interpretation capabilities of even an expert cardiologist,” said Subbu Venkatraman, chief technology officer at EKO.
Collaboration with Mayo Clinic
Berkeley, Calif.-based Eko and Mayo Clinic initially trained the algorithm on 44,959 patients to identify those with ventricular dysfunction – defined as an ejection fraction below 35% – using ECG data alone. It has been updated since to include other ECG factors. Data from a January 2019 study in Nature Medicine, one of two used to support the breakthrough designation, found the algorithm’s sensitivity and specificity were 86.3% and 85.7%, respectively, with overall accuracy of 85.7%.
With the EUA, Eko will begin offering its LVEF screening solution at Mayo Clinic before expanding the algorithm to other providers on the Eko platform.
The company won FDA approval of its atrial fibrillation and heart murmur algorithms in January and is in the early stages of deploying them to its more than 40,000 customers. “Early algorithm customers are using the tool to identify AFib and heart murmurs during in-person screening and, especially in the COVID-19 pandemic, during telehealth visits,” Bellet said.
Enabling use of the ECG algorithm with COVID-19 patients is expected to accelerate its regulatory review. “This EUA for COVID-19-related care is an important step in moving this technology closer to the front lines of care, and we look forward to continuing work with the FDA on a broader clearance of the technology,” he said.
Others using AI to improve coronavirus outcomes
Eko is one of several companies looking to apply intelligent tools to improve cardiac outcomes during the pandemic. In April, Paris-based startup Cardiologs Technologies SAS kicked off a clinical trial to assess the use of its AI platform to remotely monitor cardiac safety in COVID-19 patients being treated with the malaria drug hydroxychloroquine. The study, which is using ECG data collected from smartwatches, could help to detect and prevent serious complications linked to the drug, including QT prolongation, an electrical disturbance that can lead to severe heart arrhythmias and death.
U.K.-based Ultromics Ltd. has teamed up with Mayo Clinic for the purpose of using its AI software, Echogo Core, to analyze ECGs of COVID-19 patients for clues about how the virus affects the cardiovascular system. The aim is map cardiac features associated with the virus and improve patient triage and treatment.
Meanwhile, 1Qbit and its Canadian health care partners have received funding from the Digital Technology Supercluster, a Vancouver-based cross-industry collaboration, to speed the development of its XrAI technology. The machine learning-based clinical decision support tool is intended to improve the accuracy of chest X-ray interpretations, which could be critical in managing patient infected with COVID-19, including patients with enlarged heart due to heart failure or valve disorders.