FDA Clears AI Heart Failure Detection System

(Gerd Altmann, Pixabay. https://pixabay.com/illustrations/digitization-healthcare-health-6939536/)

6 Dec. 2022. The Food and Drug Administration authorized for marketing an algorithm-driven medical device for diagnosing a common, but difficult to detect, type of heart failure. The EchoGo Heart Failure system, made by Ultromics Ltd. in Oxford, U.K., in collaboration with the Mayo Clinic, was cleared late last month by FDA for detecting heart failure with preserved ejection fraction.

Heart failure occurs when the heart cannot pump enough blood carrying oxygen to other organs in the body. Centers for Disease Control and Prevention says some 6.2 million people in the U.S. have heart failure, with coronary artery disease, high blood pressure, diabetes, and obesity among the greatest risk factors. A conventional indicator of heart failure is reduced blood flow, or known as ejection fraction, from the heart’s left ventricle. However, Ultromics cites data showing in about half of heart failure cases, the heart still pumps seemingly normal but still not sufficient volumes of blood, called heart failure with preserved ejection fraction, or HFpEF.

The company says HFpEF is now the most common type of heart failure, but is a complex condition and difficult to detect. Mayo Clinic notes HFpEF has many contributors at both the cellular and organ level, with high blood pressure a common risk factor, found in 80 to 90 percent of cases. Because of the aging population in most parts of the world, HFpEF is becoming more common, but the complexity of the disorder makes it difficult to detect with conventional tools, including echocardiograms that use ultrasound to capture real-time images of a pumping heart.

Ultromics is a five year-old enterprise, spun-off from the lab of Paul Leeson at University of Oxford medical school that studies echocardiography, including use of artificial intelligence to interpret images of the heart. Ross Upton, a data scientist and doctoral student in Leeson’s lab, wrote algorithms to help overcome limitations of conventional echocardiogram analysis, which led to Upton and Leeson to found Ultromics, with Upton now as CEO.

Advances in heart failure treatments as well as detection

The EchoGo system reads a single two-dimensional echocardiogram file of the four-chamber heart, and uses algorithms in a convolutional neural network to analyze the image. A convolutional neural network combines image analysis with machine learning to reveal features and characteristics of the image layer by layer. Properties of each layer then further train the algorithm as more images are processed. The company cites independent data sets used to train, validate, and test its algorithms.

Ultromics says the need is increasing for better HFpEF diagnostics, with advances also taking place in treatments for the condition. The company cites sodium-glucose cotransporter 2 inhibitors, or SGLT2i drugs, first prescribed for diabetes, as having cardiovascular benefits, including fewer hospitalizations and deaths from heart failure.

“Although compelling progress has been made in the diagnosis and treatment of many forms of heart disease,” says Upton in a company statement, “heart failure hospitalizations and mortality continue to rise globally. It is now time to make a change.” Upton adds, “The technology improves the accuracy of HFpEF detection, enabling more patients to receive treatment which will reduce the significant burden on patients and health care systems alike.”

FDA authorizes the EchoGo as a software system for physicians to supplement conventional echocardiograms for detecting HFpEF. The agency cites data from a clinical trial of 1,285 patients, with EchoGo returning 535 true positives and 477 true negatives, translating to a sensitivity of 88 percent and specificity of 82 percent.

More from Science & Enterprise:

• Virtual Pathology Image Company Opens, Raises $15.2M
• FDA Clears Digital Health Monitoring System
• Grant Funds Implanted Brain-Computer Spinal Injury Device
• AI-Aided System to Identify Diabetes Complication Risks
• AI Analytics Boost Breast Cancer Assessment Diversity

We designed Science & Enterprise for busy readers including investors, researchers, entrepreneurs, and students. Except for a narrow cookies and privacy strip for first-time visitors, we have no pop-ups blocking the entire page, nor distracting animated GIF graphics. If you want to subscribe for daily email alerts, you can do that here, or find the link in the upper left-hand corner of the desktop page. The site is free, with no paywall. But, of course, donations are gratefully accepted.

*     *     *