A study for the medical potential of consumer wearables confirmed that the Apple Watch is very sensitive and can detect abnormal heartbeats with an accuracy rate of 97 percent. This study includes Apple Watch along with other Android Wear devices which applies a neural network called DeepHeart which record the heart rate to know about the atrial fibrillation which is a common heart problem that could lead to stroke.
DeepHeart study is a joint study between University of California, San Francisco’s Health eHealth Study and digital health startup Cardiogram. The study includes data from about 9,750 Cardiogram users, which contributed about 129 million heart rate measurements and 6,338 ECGs to train the DNN. DeepHeart learned to convert sensor measurements like heart rate and step counts into fibrillation. The model is capable of giving a 97 percent accuracy as compared to 12-lead ECG. The result of this study was published as “Passive Detection of Atrial Fibrillation Using a Commercially Available Smartwatch” in JAMA Cardiology on Wednesday.
According to Ballinger, DeepHeart is the first large-scale, peer-reviewed study that has appeared in a medical journal which demonstrates the consumer wearables ability to detect the condition of the health. DeepHeart’s learning algorithm and pre-training realized an increase in accuracy to similar artificial intelligence research.
The KardiaBand which cost about $199 takes about 30-second ECG reading to measure the electrical impulses. Apple Watch got an inbuilt optical heart rate sensor which helps in measuring the heart rate. Users wear Apple Watch on their left arm and place their right thumb on a sensor pad. This presses and inner electrode into their left wrist and vice-versa. In February 2018, the firm with collaboration with UCSF released a study in which it says the sensors of Apple Watch can detect early signs of diabetes with an accuracy of 85 percent.