By Peter Loftus 

Smartwatches could detect a common heart-rhythm disorder with relatively high accuracy when people were at rest, but less so when participants were physically active, a new study found.

The study, led by researchers at the University of California, San Francisco, tested the use of the heart-rate sensor and step counter built into Apple Inc.'s Apple Watch, to try to detect atrial fibrillation, a type of irregular heartbeat that can cause blood clots and strokes if left untreated. Researchers used a mobile application from Cardiogram Inc. -- which provided funding for the study -- to access the watch data, and a machine-learning network trained to analyze the data.

About 2.7 million Americans and 34 million people world-wide are estimated to have atrial fibrillation. Many patients don't experience symptoms, and it can go undetected until a complication occurs. Researchers say earlier detection could get patients on treatments like blood thinners to reduce risk of complications.

The study is part of a growing body of research testing mobile devices such as smartphones and watches as tools to monitor users' health and potentially intervene to improve outcomes. But the approach has limitations, including patient demographics that don't always represent the broader population -- partly because of the high cost of some devices. The Apple Watch costs between $249 and $399.

The new UCSF study is part of a broader project it lstarted in 2013 called Health eHeart, which is trying to harness mobile technologies to develop new strategies for preventing and managing heart disease. It has enrolled about 160,000 study participants and is aiming for up to one million people.

For the new analysis, researchers developed a type of machine learning known as a "deep neural network" to analyze the heart-rate and step-count data collected from about 6,680 of the study participants who had Apple Watches. The researchers used step-count data to assess whether increases in heart rates were normal due to physical activity or irregular.

Researchers provided Apple Watches to about 50 patients with atrial fibrillation while undergoing procedures at UCSF known as cardioversions to restore normal heart rhythms. Patients were flat on their backs and the watches collected heart-rate data while they were still experiencing irregular heartbeats, and then when normal heart rhythm was restored.

The algorithm was able to distinguish between atrial fibrillation and normal heartbeats with about 97% accuracy when compared with gold-standard electrocardiograms, according to results published online Wednesday by JAMA Cardiology.

"We've shown a proof of concept that it's very likely that these commonly worn devices will ultimately be very useful in helping to screen for atrial fibrillation," Dr. Gregory M. Marcus, a UCSF Health cardiologist and senior author of the study, said in an interview.

But the results were less impressive in a test of the algorithm in about 1,620 people using Apple Watch outside of the hospital in their everyday lives. The watch distinguished between atrial fibrillation and normal heart rhythm with about 72% accuracy, compared with participants' reports to the researchers of whether they had atrial fibrillation. That means there were a significant number of false positives and false negatives.

Researchers said it was more difficult to detect atrial fibrillation in active people because activities can vary heart rates, and the watch measured heart rates less frequently than in the hospital setting.

"The result is indeed humbling, indicating far more misclassification and lower positive predictive value compared with gold-standard" ECG's, Mintu P. Turakhia, director of the Cardiac Electrophysiology program at the Palo Alto VA Health Care System, wrote in a JAMA Cardiology editorial accompanying the study.

Consumers can't now use the Cardiogram app to detect whether they have atrial fibrillation, said Brandon Ballinger, co-founder of Cariogram. He said the company hopes to begin adding features later this year to identify people at risk of atrial fibrillation.

Apple didn't participate in the UCSF study. The company announced last year the start of its own study with Stanford University researchers to assess the ability of Apple Watch to detect atrial fibrillation.

--Tripp Mickle contributed to this article.

Write to Peter Loftus at peter.loftus@wsj.com

 

(END) Dow Jones Newswires

March 21, 2018 11:14 ET (15:14 GMT)

Copyright (c) 2018 Dow Jones & Company, Inc.
Apple (NASDAQ:AAPL)
Historical Stock Chart
From Mar 2024 to Apr 2024 Click Here for more Apple Charts.
Apple (NASDAQ:AAPL)
Historical Stock Chart
From Apr 2023 to Apr 2024 Click Here for more Apple Charts.