Wearable movement data links to hundreds of human diseases

In a cohort of 70,473 human participants followed for up to 10 years, researchers analyzed 11.8 million person-hours of wearable accelerometer data. The team derived 210 digital phenotypes covering physical activity, sedentary behavior, sleep, circadian rhythms, step counts, and step intensity. They identified 8,149 associations between these phenotypes and 406 diseases, along with 36,614 associations across 956 health-related traits. Step intensity and circadian rhythm emerged as the domains most prominently associated with diseases. An artificial intelligence framework trained on the data discriminated 112 incident diseases with an AUC above 0.7, including incident Parkinson's disease (AUC = 0.873) and dementia (AUC = 0.841). Lower disease risks were linked to benchmarks of at least 0.8 hours daily of moderate-to-vigorous physical activity, 10,000 steps daily, and 7.8 hours of daily sleep.