March 21, 2018
The diagnostic potential of wearable technology took a promising step forward this week with new evidence that Apple Watch and deep learning app Cardiogram can accurately predict atrial fibrillation.
In a peer-reviewed clinical study, results from an investigation led by UC San Francisco (UCSF) and Cardiogram demonstrated that the condition can be detected with 97 percent accuracy using a combination of artificial intelligence and preset statistical benchmarks.
Atrial fibrillation is characterized as an abnormal heart rhythm with irregular and rapid beating of the atria. The condition, which doesn't always present symptoms, is associated with an increased risk of heart failure, dementia, and stroke. Estimates from the CDC say anywhere from 2.7-6.1 million Americans have AFib, with greater prevalence among those over 65 years old.
The results of the Cardiogram study, published in JAMA Cardiology, offer hope that more of those who have the condition can be properly diagnosed and treated.
Available on Apple Watch, Android Wear and other consumer wearables, Cardiogram measures heart rate every five minutes and compiles the data to provide insights on sleep, stress and overall fitness.
The investigation included 9,750 participants enrolled in UCSF's Hearth eHealth Study. More than one hundred million heart rate and step counts were sifted by Cardiogram's DeepHeart neural network, a deep learning program model built using electrocardiograms, the current and more expensive diagnostic standard for a range of heart conditions.
A significant takeaway from the study is that the passive artificial intelligence app was more effective at detecting AFib than a dedicated EKG wristband. The use of passive sensors could drastically reduce the cost of detecting certain conditions compared to medical-grade equipment.
The study only focused on subjects with either a known risk or self-reported fibrillations, meaning further research will be necessary to confirm the effectiveness of Cardiogram in a random control trial.
Cardiogram researchers said the study's publication marks the first of its kind to demonstrate the ability of consumer wearables to detect a major health condition.