Related Experiment Video
Updated: Sep 19, 2026
Methodology for Establishing a Community-Wide Life Laboratory for Capturing Unobtrusive and Continuous Remote Activity and Health Data
Published on: July 27, 2018
The Mean and Variability of Morning Hour Activity Over 2-3 Days Predict 5-Year Mortality and Frailty Risk in a
Megan Huisingh-Scheetz1, Sylvia Brown2, Michelangelo Pagan1
1University of Chicago, Chicago, Illinois, USA.
Background:
Activity pattern features measured by continuous accelerometry are strongly associated with frailty, making accelerometry a promising, scalable screening tool for frailty. However, adherence to typical 5-7 day wear protocols decreases with age, reducing feasibility. Whether accelerometry measures collected over shorter wear periods predict frailty decline is unknown. In cross-section, we have shown that frailty is most strongly differentiated by morning activity patterns.
Objective:
The objective of this study was to determine whether two morning activity measures, mean hourly activity and within activity variance, measured over 72 h predicted 5-year mortality and frailty progression among survivors.
Methods:
We conducted a secondary data analysis of the 2010-2011 accelerometry substudy and 2015-2016 data of the National Social Life, Health and Aging Project which included n = 584 community-dwelling older adults; n = 104 died at 5 years. Wrist accelerometers were worn for 72 h in 2010-2011, and an adapted frailty phenotype was measured in both years. We employed a mixed effects location scale model to calculate mean morning (7:00-11:59 am) hourly activity (counts per minute z-score) and within-subject morning hourly activity variance (z-score) for each participant, adjusted for day of week and month of wear. Second, 5-year mortality (logistic) and adapted frailty phenotype scores (range 0-4, ordinal) were regressed on baseline frailty and both accelerometry z-scores in a 2-stage model, adjusting for covariates.
Results:
After adjusting for baseline frailty, demographics, comorbidities, BMI, and cognition, higher mean morning hourly activity predicted lower 5-year mortality risk (OR = 0.64 for 1 SD increase, p = 0.04). Conversely, higher morning activity WS variance predicted better frailty scores among survivors (OR = 0.67 for 1 SD increase, p = 0.01).
Conclusions:
Higher activity levels and higher day-to-day morning activity variance measured over just two to three mornings differentiate those at higher risk of death and frailty decline, increasing the feasibility of using accelerometry as a digital frailty screening tool.
