“We’ve called it Shazam for heartbeats. You know Shazam, the music identification app that helps you figure out what song is playing in Starbucks. We do the same thing, but for heart sounds.”
That’s the way Eko CEO Connor Landgraf described his company’s new product in a recent interview. Last week, the company — best known for its digital stethoscopes — launched the Sensora platform, which consists of AI-powered software designed for the earlier detection of cardiac disease.
Eko’s digital stethoscopes have been on the market for nearly a decade, and more than 300,000 clinicians use them, Landgraf said. The Sensora platform works by connecting to those digital stethoscopes and identifying structural murmurs, a sign of valvular heart disease.
The platform uses a machine learning algorithm to search through its database of heart sounds and determine whether the sounds that a primary care provider hears through their stethoscope are normal or abnormal.
“Instead of finding the song that’s playing, we determine whether there is heart disease or not,” Landgraf declared. “Then we provide that notification to the clinician with an app that runs on a smartphone or tablet. The physician gets that interpretation right there in the exam room, and they then make their decision about next steps for the patient.”
Eko is selling the Sensora platform primarily to organizations that treat patients “from beginning to end,” such as large health systems, Landgraf explained. He said this is because the software could identify that a patient requires specialty follow-up care, and these organizations would be able to easily refer them to a practitioner within their enterprise.
The Sensora platform was cleared by the FDA last summer, and it is in its “early access deployment” stage, Landgraf said. He didn’t disclose which health systems are currently using the product or the price of the software’s installment.
Eko isn’t the only company developing AI technology for the earlier detection of heart disease. There’s plenty of companies in that space, such as Cardiologs and Caption Health (which GE HealthCare acquired this month). Eko differentiates itself by focusing specifically on heart sounds, Landgraf declared.
“What’s unique about heart sounds is that the data is super easy to collect. I can guarantee that almost every single person on the planet who has gone to a doctor’s office has had the doctor listen to their heart sounds and lung sounds. The accessibility and ubiquity of heart sounds is so powerful, and then we add in the machine learning accuracy,” he said.
To collect the data that informs the Sensora platform, Eko partnered with health systems for prospective data collection studies. The company collected heart sound data from hundreds of thousands of patients with different types of cardiovascular disease, along with diagnostic reports.
During this process, Eko made sure that the data it collected was diverse in terms of age, gender and race, Landgraf noted.
The company also measured the performance of its AI software against the expertise of general practitioners. The study concluded that the Sensora platform more than doubles primary care physicians’ likelihood of detecting significant valvular heart disease.
“We were able to see a very substantial increase in diagnostic accuracy, specifically around sensitivity. If a physician hears something, they’re pretty good at knowing what they hear. If they hear a murmur, most of the time they are correct and they did hear that. But there are still a lot of cases where there are murmurs but they’re very subtle, and the physician doesn’t necessarily hear it. So what we can provide is additional accuracy, or what we call sensitivity, which is making sure that physicians didn’t miss any patients who have just more subtle signs of disease,” Landgraf explained.
Use of the Sensora platform is not reimbursed by payers. While there are currently no CPT codes available for the AI software, Landgraf said that is something that the company is “looking to advocate for long-term.”
Photo: Eko