The Apple Watch may one day notify users of a heart irregularity that can lead a to a stroke. But new research shows the technology isn’t quite there yet.
In a study published in JAMA Cardiology, researchers at the University of California, San Francisco (UCSF) found that the Apple Watch, combined with an algorithm designed to detect atrial fibrillation (AF), performed well among sedentary patients undergoing a medical procedure. Researchers showed the Apple Watch-algorithm combination was 98% accurate among a group of 51 participants undergoing cardioversion, a corrective procedure that uses electrical shocks to restore irregular heart rhythms back normal.
RELATED: Study shows wearables lack clinical impact, but researchers haven’t lost hope
But among an ambulatory group of more than 1,600 participants, the watch and algorithm were just 68% accurate.
On one hand, researchers said the study demonstrated “a commercially available smartwatch can passively detect AF” using a mobile application and machine learning. But they added the experiment highlights the challenges of using the approach for “constantly mobile individuals in natural environments.”
Ultimately, they said more research is needed to identify the “optimal role” for smartwatches to identify or predict cardiac irregularities.
“This is the first heads-up: your smartwatches have the capability of doing this, so it’s coming and it’s theoretically possible,” Gregory Marcus, a USCF cardiologist who led the study told the Washington Post.
Failure of a deep learning #AI algorithm to detect atrial fibrillation via smartwatch prospectively @JAMACardio @UCSF @GeoffTison https://t.co/vvU23bStKZ pic.twitter.com/lw0RV6AkGi
— Eric Topol (@EricTopol) March 21, 2018
In accompanying editorial, Stanford University cardiology researcher Mintu P. Turakhia, M.D., called the results “humbling" and urged the industry to “think creatively but prudently” about how new technology is used to predict disease.
“With computational advances and more training data, it is possible that these algorithms may improve,” he wrote. “However, there is also the possibility that they hit a performance ceiling that remains inferior to an accepted gold-standard. What, then, should be the tradeoff that we are willing to accept between high diagnostic accuracy and convenience, ubiquitousness, and continuous monitoring?”
Research into this intersection of technology and medicine isn’t slowing down anytime soon. Late last year, Stanford partnered with Apple and American Well to launch a similar study exploring how the Apple Watch can identify users with AF and direct them to a physician.
In an interview with FierceHealthcare, Stanford Medicine Dean Lloyd Minor, M.D., said he has high hopes for the study.
“The goal is to provide the most accurate information possible and we believe this study is going to be able to do that,” he said.