The first artificial intelligence-enabled cardiac ultrasound software has received De Novo marketing authorization, the FDA announced today. Developed by Caption Health, the Caption Guidance tool guides medical professionals through a heart disease diagnostic test that normally requires an ultrasound expert.

“Echocardiograms are one of the most widely-used diagnostic tools in the diagnosis and treatment of heart disease,” Robert Ochs, deputy director of the Office of In Vitro Diagnostics and Radiological Health in the FDA’s Center for Devices and Radiological Health, said in a statement. “Today’s marketing authorization enables medical professionals who may not be experts in ultrasonography, such as a registered nurse in a family care clinic or others, to use this tool.”

According to the agency, the software is now indicated for use in adult patients for the acquisition of standard views of the heart from different angles. It’s currently compatible with a Teratech Corporation-produced diagnostic ultrasound system, and could be used with other ultrasound imaging systems with similar specifications.

According to Caption Guidance, the tool guides practitioners in real time on how to place and move the scanner to produce the best quality image. The software records and stores the highest quality images from each scanning session, and assists analysis with automatic automatic calculation of the ejection fraction — although the final assessment and diagnosis is still conducted by the cardiologist.

The agency said that it reviewed data from two independent studies of the software. The first of these had 50 sonographers scan patients with and without the tool, and found that both settings yielded comparable image quality. The second had eight registered nurses use the software to collect echocardiography images, the quality of which were then assessed by five cardiologists who found them to be of appropriate diagnostic quality.

WHAT’S THE IMPACT

Heart diseases are the leading killer in the US, and according to the CDC account for one in every four deaths among men and women. Early detection is the first step in preventing serious episodes or managing a long-term condition, meaning that a tool expanding the number of medical staff able to initiate diagnostic imaging could go a long way toward cutting down on new cases.

Further, Caption Health’s De Novo is the latest in a growing number of AI-based tools being cleared to scale diagnostic care and reduce the burden among overworked specialists.

“This is especially important because it demonstrates the potential for [AI] and machine learning technologies to increase access to safe and effective cardiac diagnostics that can be life-saving for patients,” Ochs said.

THE LARGER TREND

The past few years has seen the FDA clear dozens of algorithms and clinical decision support (CDS) tools, having just recently released a revised version of its draft guidance on the latter in September. Within the area of cardiovascular diagnostics specifically, health technology firm Ultromics received a 510(k) for its EchoGo Core system in November while digital stethoscope maker Eko picked up its own clearance for an atrial fibrillation and heart murmur detection algorithm just a few weeks ago.

Last year, Apple announced the results of the Apple Heart Study, an ambitious digitally-enabled research project that also tapped the talents of Stanford University, BioTelemetry, and telemedicine company American Well. Through the study, more than 400,000 individuals helped Apple and Stanford learn about atrial fibrillation, and the predictive and therapeutic value of the Apple Watch’s algorithms.

At HIMSS20 next month, American Well General Manager for Strategic Solutions Dr. Jeff Kosowsky and Stanford Associate Professor of Medicine Dr. Mintu Turakhia will discuss lessons learned from the study and how the landmark study could have repercussions much bigger than a consumer wearable.

“What the New England Journal was all about was really a cardiology study with a little footnote about ow this was done virtually with an Apple Watch,” Kosowsky told MobiHealthNews. “But what’s really unique here is how this has been done differently from other studies. That’s what was interesting in the first place to the FDA, to the four parties involved, and to the industry. So what we’re going to be talking about here is what I think is the real new part, the story that wasn’t told.”

The Apple Heart Study demonstrated a proof of concept for a new kind of research study, Kosowsky said. Not just a traditional trial with digital add-ons, but a new kind of remote trial built from the ground up.

“This is not an angle of how a CRO becomes digital,” he said. “It’s really more of an angle about how consumer companies take competencies that are more in the clinical or consumer space and creates a new notion of the clinical study, one that can have 400,000 participants. One where the actual people from Stanford Cardiology never spoke to a single patient.”

So what were the lessons learned from this new experience? Kosowsky cited four. 

One, organizations should emphasize and prioritize collaboration, and recognize ahead of time that research institutions, consumer companies, and digital health firms all have different ways of operating and it may not be natural or easy to synthesize them together.

Two, when dealing with such large numbers of patients, it’s important to automate as many of the processes involved as possible.

Three, Kosowsky said, organizations should shift their thinking from enrollment to a mindset of consumer engagement. That’s what he chalks up Apple’s success in recruiting so many participants to.

Finally, in the model they used it’s important to make sure providers are engaged as well, and give the study as much attention as their other duties.

“The provider perspective is really important because one of the ways we made this work is sort of brains and muscles,” Kosowsky explained. “The Stanford providers were the brains, but they never spoke to any patients. The muscles were providers from our network that we got to recruit. So the patient before may have been someone with a headache and the person after may have been a person with an allergy and they are providing urgent care/primary care, so we have to have also the provider experience, which means we have to make sure the providers are motivated and don’t see this as an undue burden.”

Kosowsky and Turakhia will present “Digital Models and Telehealth for Clinical Trials Evidence” Wednesday, March 11 at 1:00 PM in Room W414D.