Skip to main content

Apple Watch detects heart irregularity with 97 percent accuracy

Apple Watch detects heart irregularity with 97 percent accuracy


Not good enough for official diagnosis, but useful as a screening tool

Share this story

Photo: Courtesy of Cardiogram

An Apple Watch app can detect a common form of heart irregularity with 97 percent accuracy, according to results presented at the Heart Rhythm Society last week. This is promising because people with the heart condition — called atrial fibrillation — often have no symptoms, making screening and diagnosis difficult.

Cardiogram — a heart rate monitoring app used with the Apple Watch — worked with the University of California, San Francisco on its Health eHeart Study. From March 2016 to about February of this year, Cardiogram recruited over 6,000 participants. Though most participants had normal readings, about 200 had already been diagnosed with atrial fibrillation, or afib. Cardiogram engineers then used this data to train an algorithm to predict afib, says co-founder Brandon Ballinger.

Atrial fibrillation is a form of irregular heartbeat that happens when the two upper chambers of the heart don’t beat in sync with the two lower chambers. The condition can increase the risk of everything from heart attack to kidney disease to dementia, says Greg Marcus, a professor at UCSF who runs the Health eHeart Study. Symptoms include being short of breath and having heart palpitations, but sometimes people have no symptoms at all.

This makes treatment difficult. And even for the people who do get symptoms, they might not have them all the time, so it’s hard to detect the irregularities. These patients will go to the doctor, who will then send them with various monitoring sensors, like the Zio patch or LifeWatch, to wear continuously. But these work for only a couple weeks, according to Marcus. Another option is implantable devices, like Medtronic’s Linq, but these are invasive.

The advantage of the Apple Watch with Cardiogram is that it monitors the user continuously without the requiring the user to do anything, and it’s included in a watch that the user already owns. That said, the results are only a first step. “I think it’s unlikely that, at least in the next few years, this sort of algorithm is sufficient to make a diagnosis of atrial fibrillation,” says Marcus. “It will be very useful to screen, but the diagnosis will still require confirmation using a conventional EKG.”

Next, Ballinger and the Cardiogram team are interested in using heart rate data to detect other forms of disease. “There’s a little evidence that even areas like diabetes can show up in heart rate data and there are other conditions, too,” says Ballinger. “The interesting thing about the heart is, because it’s connected to the autonomic nervous system, it provides a window into your entire internal organ system, so the sky’s the limit.”