A study from health startup Cardiogram and the University of California argues that wearables such as Apple Watch can help detect common, serious conditions like sleep apnea and high blood pressure.
Cardiogram and UC previously reported that Apple Watch could measure irregular heart rhythm with about 97% accuracy. The new work puts sleep apnea detection around 90% and high blood pressure around 82%. Being able to surface those signals from everyday wear would be a meaningful shift for personal health screening.
Deep learning in the loop
About 6,115 participants used the Cardiogram app with Apple Watch for a period. A deep learning model called DeepHeart flagged sleep apnea in 1,016 of them and high blood pressure in 2,230. Co-founder Brandon Ballinger said DeepHeart was trained on data from 70% of participants and tested on the remaining 30%. Co-founder Johnson Hsieh added that while the study used Apple Watch, similar results should be possible on any wearable with a heart-rate sensor.
After Google Brain’s diabetic eye disease results in December 2016 and Stanford’s skin cancer work in January 2017, the Cardiogram and UC project ranks among the larger deep learning efforts applied to health.
