Circadian rhythm analysis using wearable-based accelerometry as a digital biomarker of aging and healthspan

Recognizing the pivotal role of circadian rhythm in the human aging process and its scalability through wearables, we introduce CosinorAge, a digital biomarker of aging developed from wearable-derived circadian rhythmicity from 80,000 midlife and older adults in the UK and US. A one-year increase in CosinorAge corresponded to 8–12% higher all-cause and cause-specific mortality risks and 3–14% increased prospective incidences of age-related diseases. CosinorAge also captured a non-linear decline in resilience and physical functioning, evidenced by an 8–33% reduction in self-rated health and a 3–23% decline in health-related quality of life score, adjusting for covariates and multiple testing. The associations were robust in sensitivity analyses and external validation using an independent cohort from a disparate geographical region using a different wearable device. Moreover, we illustrated a heterogeneous impact of circadian parameters associated with biological aging, with young (<45 years) and fast agers experiencing a substantially delayed acrophase with a 25-minute difference in peak timing compared to slow agers, diminishing to a 7-minute difference in older adults (>65 years). We demonstrated a significant enhancement in the predictive performance when integrating circadian rhythmicity in the estimation of biological aging over physical activity. Our findings underscore CosinorAge’s potential as a scalable, economic, and digital solution for promoting healthy longevity, elucidating the critical and multifaceted circadian rhythmicity in aging processes. Consequently, our research contributes to advancing preventive measures in digital medicine.


Self-Rated Health
(higher = poorer health)

EQ-5D
(lower = poorer HRQoL) Generalized linear models were adjusted for age, ethnicity/race, Townsend deprivation index (UKB only), baseline assessment center (UKB only), education, employment, body mass index, smoking status, alcohol consumption, and comorbidities.We account for complex survey design in NHANES.CAA CosinorAgeAdvance, SE standard errors, EQ-5D European Quality of Life-5 Dimensions 5-levels, ADL Activities of Daily Living, HRQoL Health-related quality of life.*p values < 0.05.We standardized the CAAs and outcomes to have a mean value of 0 and a standard deviation of 1.

Supplementary Fig. 1
Circadian rhythm parameters with chronological age.a Midline Estimating Statistic of Rhythm (MESOR).b Amplitude.c Acrophase.Mean values and the standard errors are shown for individuals categorized by chronological age.

Supplementary Fig. 2
Associations between CosinorAge and established biological aging measures in validation cohorts stratified by sex.a,b Correlations with chronological age, Klemera-Doubal biological age (KDM BA), and phenotypic age (PhenoAge) in the UKB internal validation cohort stratified by sex.c,d Correlations with chronological age, KDM BA, and PhenoAge among females in the NHANES external validation cohort stratified by sex.r Pearson's correlation coefficient.

Table 1 .
Gompertz proportional hazards regression models for CosinorAge using the training dataset.

Table 2 .
Sex-stratified associations between CosinorAge advancement and mortality risks.

Table 3 .
Sex-stratified associations between CosinorAge advancement and disease incidences.
*Cox proportional hazard regression models were adjusted for age, ethnicity/race, Townsend deprivation index, baseline assessment center, education, employment, body mass index, smoking status, and alcohol consumption.Diabetes-and chronic respiratory disease-specific incidences were adjusted for age to achieve model convergence.Participants with prevalent diseases were excluded from the analyses.First incidence was defined as the first occurrence of any of the aforementioned diseases.*p<0.05.Supplementary

Table 4 .
Sex-stratified associations between CosinorAge advancement and age-related functional performances.

Table 5 .
Sex-stratified age-dependent variations in circadian rhythm metrics by CosinorAge advancement.Cox proportional hazard regression models were adjusted for age, ethnicity/race, Townsend deprivation index (UKB only), baseline assessment center (UKB only), education, employment, body mass index, smoking status, alcohol consumption, and comorbidities.Diabetes-and chronic respiratory disease-specific mortalities were adjusted for age to achieve model convergence.We account for complex survey design in NHANES.Participants with prevalent diseases corresponding to the related disease-specific mortalities were excluded from the relevant analyses.Agingrelated mortality is defined as death resulting from any of the following diseases: CVD, cancer, diabetes, neurodegenerative (UKB only), and chronic respiratory diseases.*p<0.05.Model 1 excluded participants with shift work history (UKB only).Model 2 additionally adjusted for sleep disorders.Model 3 additionally adjusted for circadian rhythm fragmentation and stability.Cox proportional hazard regression models were adjusted for age, ethnicity/race, Townsend deprivation index, baseline assessment center, education, employment, body mass index, smoking status, and alcohol consumption.Diabetes-and chronic respiratory disease-specific incidences were adjusted for age to achieve model convergence.Participants with prevalent diseases were excluded from the analyses.First incidence was defined as the first occurrence of any of the aforementioned diseases.*p<0.05.Model 1 excluded participants with shift work history.Model 2 additionally adjusted for sleep disorders.Model 3 additionally adjusted for circadian rhythm stability and fragmentation.Model 4 excluded participants with less than 2 years of follow-up time.

Table 8 .
Sensitivity analyses of associations between CosinorAge advancement and age-related functional performances.Generalized linear models were adjusted for age, ethnicity/race, Townsend deprivation index (UKB only), baseline assessment center (UKB only), education, employment, body mass index, smoking status, alcohol consumption, and comorbidities.We account for complex survey design in NHANES.CAA CosinorAgeAdvance, SE standard errors, EQ-5D European Quality of Life-5 Dimensions 5-levels, ADL Activities of Daily Living, HRQoL Health-related quality of life.*p values < 0.05.We standardized the CAAs and outcomes to have a mean value of 0 and a standard deviation of 1. Model 1 excluded participants with shift work history (UKB only).Model 2 additionally adjusted for sleep disorders.Model 3 additionally adjusted for circadian rhythm stability and fragmentation.