[Skip to Content]
[Skip to Content Landing]
Views 2,432
Citations 0
Biotech Innovations
January 14, 2020

Studying New Approaches to Detect Atrial Fibrillation

JAMA. 2020;323(2):111. doi:10.1001/jama.2019.21593

Researchers tested a smartwatch-based irregular pulse notification algorithm for suspected atrial fibrillation among more than 400 000 US participants in a recent trial. The findings, published in the New England Journal of Medicine, came from the Apple Heart Study, a collaboration between technology giant Apple and Stanford Medicine.

Over a median of 117 days, 2161 participants received irregular pulse notifications identifying possible atrial fibrillation and were then mailed an electrocardiography (ECG) patch to wear for 7 days. Thirty-four percent of participants who mailed back patches had the condition on subsequent ECG readings. The ECGs confirmed irregular pulse notifications as atrial fibrillation 84% of the time among participants who used the watch and the patch simultaneously.

However, only a fifth of participants with irregular pulse notifications returned an ECG patch for analysis. The authors of an accompanying editorial wrote that while wearable devices will facilitate health research, “obtaining long-term participant commitment and compliance may become a greater challenge.”

The study used the Apple Watch’s light-emitting and light-sensitive technology to measure changes in blood flow through the skin, which correlate with pulse and heart rate. Using the same basic approach, a separate team of researchers recently published a small proof-of-concept study in JAMA Cardiology looking at whether a pretrained neural network could classify atrial fibrillation in digital camera–recorded videos 5 people at a time, a potentially time-saving and cost-efficient approach.

The high-throughput technique correctly identified heart rhythms in about 80% of 64 videos. With further development, the method could be adapted for automated triage in waiting rooms or mass casualty settings, although facial recognition’s privacy issues must be addressed, the journal’s editors wrote in an accompanying letter.