This article is more than 5 years old. Information may no longer be current.
AI-based algorithms cleared by FDA for AF, valvular disease detection
Eko announced the FDA cleared its cloud-based artificial intelligence algorithms for use with the company’s digital stethoscopes for the detection of murmurs, atrial fibrillation, tachycardia, bradycardia and measurements of QRS duration.
The company reported that its AI can identify heart murmurs, a symptom of valvular heart disease, with 87% sensitivity and 87% specificity, compared with 43% sensitivity and 69% specificity for traditional stethoscopes.
“When you think about the stethoscope in the hand of the average physician, the specificity and sensitivity of being able to hear murmurs is significantly increased by having AI associated with the stethoscope,” Ami B. Bhatt, MD, FACC, director of the adult congenital heart disease program and director of outpatient cardiology at the Corrigan Minehan Heart Center at Massachusetts General Hospital and associate professor in medicine at Harvard Medical School, told Healio. “What I love is that we can still rely on the clinician’s judgment and the clinical document, but we're simply enhancing that by using these AI algorithms.”
In addition, the algorithms may detect AF with 99% sensitivity and 97% specificity when analyzing the 1-lead ECG tracing from the company’s digital stethoscope products, enabling providers to screen patients for it during a standard physical exam, according to the release.
“I use it in my practice for a few different reasons. I use it to teach my students and oftentimes I'm listening to a murmur and then asking them to listen to a murmur. Then, I'm hoping that by our words, we can determine if we're hearing the same thing,” Bhatt said in an interview. “It's also recording device that we can record murmurs with and talk about them later. For the patients, my hope is that patients that have their own murmur can then go to someone who doesn't know them as well as me, their cardiologist, and ask if they are hearing and seeing the same thing.”
Using signal processing and convolutional neural networks, the cloud-based software analyzes ECG and heart sound/phonocardiogram data to interpret the acquired signals, according to the release.
The company also stated that:
- The AI is intended to provide physician support for the evaluation of heart sounds and ECGs;
- It does not differentiate different types of murmurs or identify other arrhythmias;
- It is not intended as a sole means of diagnosis; and
- The assessments provided by the AI are significant only when used with physician over-read and when used on adults.
“The future of outpatient cardiology is in following the patient journey and providing cardiac care to the patients in the communities where they live,” Bhatt told Healio. “One problem for cardiologists is to not be able to have a physical exam with the patient when they're not coming into a major academic medical center. Devices like this allow us to have, not only a potential cardiac exam, but that correlating with a single-lead ECG, makes it much more likely that we're going to be able to deliver advanced cardiac care in the communities where people live and help the local physicians provide high-level care.”
Disclosures: Bhatt reports she is a scientific advisor for Eko.