11 million days of longitudinal Fitbit data reveal novel future health insights

Abstract Wearable devices offer a scalable, objective measurement of physical activity, but most studies reduce these data to a single metric over 7-days. Here, we compare wearable-derived phenotyping approaches to understand their impact on activity-disease associations, analyzing 11 million days of Fitbit data from 29,303 All of Us Research Program participants. We derive four daily activity metrics (step count, peak 1-min cadence, peak 30-min cadence, and heart rate per step) across five time-windows (1-day, 1-week, 1-month, 6-months, 1-year), then run phenome-wide analyses across > 700 incident and >1,400 prevalent disease outcomes. Though all phenotypes are highly correlated, longer measurement windows yield stronger, more stable associations: 1-year step count was associated with 389 prevalent and 61 incident outcomes (versus 271 and 33 for 1-day step count) after Bonferroni-correction, including novel links to chronic pain syndrome and SARS-CoV-2. Differences between prevalent and incident associations suggest that activity metrics may act as both early markers of disease and risk factors. These findings highlight how large-scale, longitudinal wearable data can advance understanding of health and disease.