By Diagnostics World Staff
November 13, 2018 | AliveCor has announced research suggesting the use of AliveCor’s AI technology as a potential alternative to surgically implanted heart monitors.
In a study presented at the American Heart Association Scientific Sessions, researchers from Northwestern Medicine demonstrated that the performance of AliveCor’s KardiaBand with an experimental version of SmartRhythm artificial intelligence (AI) in detecting atrial fibrillation (AFib) compares favorably to an implantable loop recorder (ILR), the current industry standard.
Traditionally, ILRs have been viewed as the industry gold standard, allowing for long-term surveillance of AFib. ILRs, however, require surgical implantation; can cost over $20,000 to be implanted and monitored; and provide no real-time feedback to the patient. In contrast, KardiaBand allows patients to record and see a lead I ECG from the wrist, with instant analysis for AFib. With the SmartRhythm AI, which continuously evaluates the smartwatch-provided heart rate and activity level, users are prompted to record an ECG when discordance between the two is detected.
The Food and Drug Administration cleared KardiaBand in the U.S. last November. The band, made for the Apple Watch, includes an embedded electrocardiogram (EKG or ECG) device. KardiaBand lets users capture a medical EKG and quickly identify abnormal heart rhythms such as atrial fibrillation (AF).
SmartRhythm is a feature within the KardiaBand app, an artificial intelligence platform that uses data from Apple Watch’s heart rate and activity sensors to evaluate the correlation between heart activity and physical activity. When the two don’t align, the app prompts users to take an immediate EKG.
“We built a deep neural network—basically artificial intelligence using an autoregressive neural network—it’s a trained, unsupervised neural network where we trained it on normal people—healthy people—and people that had arrhythmias,” Vic Gundotra, CEO of AliveCor, told Diagnostics World a year ago. “We asked the artificial intelligence to make a prediction based on your heart rate right now: What should your heart rate look like in the next minute?”
Investigators at Northwestern Medicine sought to determine whether KardiaBand with an experimental version of SmartRhythm could accurately detect episodes of AFib compared to an ILR.
Researchers began the study in May of 2017. In this cohort, over 31,000 hours of the experimental SmartRhythm and ILR rhythm monitoring were collected over a four-month period and were analyzed to determine comparative efficacy. Patients with a previously implanted ILR device and a history of paroxysmal AF were eligible for enrollment.
They found that KardiaBand ECG prompted by the experimental SmartRhythm detected 74 of 76 episodes of AFib of at least 1 hour in duration, with a sensitivity of 97.4%. The correlation for AF episode duration between KardiaBand and ILR was high, at 0.997. The investigators concluded that KardiaBand with the experimental SmartRhythm holds promise as an inexpensive and non-invasive approach to long term AFib surveillance and management. The studied method used to detect AFib is investigational and pending FDA review.
“As atrial fibrillation is increasingly recognized as a major cause of stroke, an inexpensive and accurate wearable device will be an important tool for detecting and managing heart arrhythmias in millions of individuals who either already have the disease or are at risk of developing atrial fibrillation,” said senior author Rod S. Passman, MD, MSCE, in a company press release. “Our study is a proof-of-concept that you can use a wearable device to not only show whether or not you have atrial fibrillation, but how much atrial fibrillation you are having. We are excited for the impact that wearable technologies will have on the future of care.”
“This new research validates KardiaBand performance with our experimental AI algorithms against legacy solutions and gives patients an inexpensive and immensely more comfortable way to allow long term heart monitoring,” Gundotra said in the same statement. “With cost-savings in the tens of thousands of dollars, we’re confident that future versions will have the ability to lead the next generation of monitoring technology."