Eko's algorithm can help non-cardiologists detect heart murmurs in seconds
Non-cardiologists may soon be able to detect heart murmurs with greater confidence thanks to an algorithm currently in development.
This is what cardiac monitoring enterprise Eko suggested this weekend at the American Heart Association’s Scientific Sessions 2018 conference, arguing that their solution may potentially bridge the clinical skill gap between the 27,000 U.S.-based cardiologists and 3.8 million other types of clinicians in detecting such events with stethoscopes. The claim is based on the findings of a clinical study in which the algorithm outperformed a majority of radiologists in identifying such instances.
“Detecting heart murmurs via a stethoscope is inherently challenging. The differences between normal and healthy heart sounds can be very hard to detect and are made more challenging by different heart rates,” Connor Landgraf, CEO of Eko, told HCB News. “Most practitioners who don't have experience listening to thousands of heart sounds are not confident in their abilities. Our goal is to empower every physician and clinician to confidently screen patients for heart diseases.”
Though cardiologists effectively diagnose 90 percent of cardiac events using a stethoscope, internal medicine and family practice physician residents have been found to misdiagnose these conditions 80 percent of the time, according to a study published in the Journal of the American Medical Association.
The neural network AI algorithm is trained on thousands of algorithms and when paired with the FDA-cleared Eko Core and Eko Duo devices, allows any and all clinicians to more accurately assess heart murmurs in a few seconds.
The solution was shown to fare better than a majority of five cardiologists in an evaluation of an independent data set of pediatric heart sounds, and was also tested against echocardiogram imagery, considered to be the gold standard for diagnosing structural heart conditions.
Access and use of echocardiograms can be a challenge due to their cost and length of time for exams, ushering in the need for Eko’s algorithm, according to Landgraf.
“Most patients will never receive an echocardiogram, and so early screening for heart disease through sound is an exciting promise,” he said. “The Eko AI can enable any physician to be able to detect heart murmurs with a high level of confidence. We can combine the knowledge from tens of thousands of previous heart sound recordings to make a powerful detection tool.”
The company is currently pursuing FDA clearance for the algorithm with plans to roll it out with existing cardiac monitoring devices once approved.
