The Apple Watch is 97% accurate at detecting common abnormal heart rhythms, according to a study by the University of California, San Francisco. Heartbeat measurement app Cardiogram began the study with UCSF last year to determine whether or not the wearable could detect an oncoming stroke.
The study consisted of 6,158 participants, most of which had normal EKG readings and 200 of which had been diagnosed with paroxysmal atrial fibrillation (or an abnormal heartbeat). Engineers trained a deep neural network to identify the abnormal heart rhythms from Apple Watch heart rate data.
Testing their findings against 51 in-hospital cardioversions (a procedure that restores the heart’s normal rhythm), the team says its neural network correctly identified irregular heart activity with a 97% accuracy rate. The results hold promise for the long-running effort to detect and prevent strokes in the future.
Atrial fibrillation, the most common abnormal hearth rhythm, is believed to cause 1 in 4 strokes. Cardiogram co-founder Brandon Ballinger says two-thirds of these types of strokes can be prevented with inexpensive drugs. The team plans to continue its eHealth study and further validate its neural network.