Fitbit activity rhythms tracked with biological-aging trajectories in All of Us study

What happened

Researchers analyzing several years of Fitbit and clinical data from the National Institutes of Health’s All of Us Research Program found that daily rest-activity patterns were associated with trajectories of a blood-based biological-age measure. The peer-reviewed study, published August 22 in Nature Communications, examined 2,222 adults contributing 8,447 person-years of data.

The central finding was that the intensity, timing and stability of participants’ activity rhythms carried information about whether their biomarker-derived age appeared to accelerate over time. Higher rhythm intensity was associated with 26% to 46% lower odds of accelerated aging across the study’s analyses. Associations involving rhythm timing and regularity were stronger among female participants, while the researchers reported a more complex pattern of instability among males.

Those results make continuous wearable data potentially useful for aging research. They do not establish that a Fitbit can measure a person’s biological age or that deliberately changing an activity schedule will alter aging.

How the study worked

The investigators linked intraday step data shared from participants’ personal Fitbit accounts with electronic health-record measurements used to calculate PhenoAge. PhenoAge combines chronological age with clinical chemistry and blood-cell markers to estimate mortality-related biological risk; it is a research construct rather than a diagnosis.

Participants needed at least two PhenoAge measurements aligned with years of valid Fitbit monitoring. For an individual year to qualify, the analysis required a complete 12-month period of step data. The researchers then characterized multiple dimensions of daily rest-activity behavior, including overall activity intensity, the timing of activity and the consistency of patterns across days.

This longitudinal structure is stronger than taking a one-time snapshot: it allowed the investigators to compare wearable patterns with biological-age trajectories across multiple years. The endpoint, however, remained a biomarker classification—not disability-free survival, disease incidence or lifespan.

What the association may mean

A robust daily activity rhythm could reflect several overlapping influences, including greater physical activity, more regular sleep and work schedules, better underlying health or fewer mobility limitations. A step-count pattern cannot cleanly distinguish an internal circadian signal from behavior imposed by employment, caregiving, illness or the surrounding environment.

That ambiguity matters. The study supports the possibility that consumer wearables could provide scalable digital markers for population research or future risk-assessment models. It does not show which component of the observed rhythm, if any, is causally responsible for the PhenoAge association.

The sex-specific results also require replication. They may represent biological differences, behavioral patterns, sample composition or model sensitivity; the analysis was not an intervention designed to test those explanations.

Important limitations

Participation was highly selective. All of Us participants had to own or use a compatible Fitbit, consent to sharing its data, contribute sufficiently complete records and have repeated clinical laboratory measurements. People able to supply uninterrupted year-long wearable data may differ systematically from the broader population, limiting generalizability.

Consumer-device data introduce additional uncertainty. All of Us notes that Fitbit information is delivered through the company’s web interface without additional program-level cleaning, and a separate characterization of the program’s wearable dataset cautions that reliability can vary by metric, device type and population. Missing wear time, device changes and mobility-related step-count errors could affect derived patterns.

Most importantly, this was an observational analysis. Residual confounding and reverse causation remain plausible: healthier people may maintain stronger activity rhythms, rather than stronger rhythms producing healthier aging. No treatment, behavior or schedule was randomly assigned.

Practical meaning

The study advances a promising measurement strategy: combining dense, real-world activity records with repeated clinical biomarkers may reveal aging trajectories that occasional clinic visits miss. Before such measures can inform clinical decisions, they need independent validation, standardized processing, prospective testing and evidence that they improve meaningful health outcomes beyond established risk factors. For now, the findings are evidence for a research biomarker—not a consumer longevity score.