JMIR Publications

ABSTRACT

Background:

Although accelerometers are present in popular consumer wearable devices, to our knowledge, no prior study has demonstrated the feasibility and acceptability of accessing sufficient consumer wearable accelerometer data to compute 24-hour sleep/wake rhythm measures. Unlocking these data has the potential to support a new generation of studies with real-world implications (e.g., screening for the 24-hour sleep-wake risk factors established in research).

Objective:

We aimed to: (a) establish the feasibility of characterizing 24-hour sleep/wake rhythm measures using accelerometer data gathered from the Apple Watch in older adults with and without Mild Cognitive Impairment; and (b) examine correlations of these sleep/wake rhythm measures with neuropsychological test performance.

Methods:

Of 40 adults enrolled (mean age=67.2, standard deviation (SD)=8.4; 72.5% female), 19 had MCI and 21 had no cognitive disorder (NCD). Participants were provided devices, oriented to the study software (myRhythmWatch or myRW), and asked to use the system for a week. The primary feasibility outcome was whether participants collected enough data to assess 24-hour sleep/wake rhythm measures (i.e., ≥3 valid contiguous days). We extracted standard nonparametric and extended-cosine based sleep/wake rhythm metrics. Neuropsychological tests gauged immediate and delayed memory (Hopkins Verbal Learning Test) as well as processing speed and set-shifting (Oral Trails Parts A and B).

Results:

Using the Apple Watch-based myRW system to gather raw accelerometer data is feasible in older adults with and without MCI. Sleep/wake rhythms variables generated from this system correlated with cognitive function, suggesting future studies can use this approach to evaluate novel, scalable, risk factor characterization and targeted therapy approaches.

Conclusions:

Using the Apple Watch-based myRW system to gather raw accelerometer data is feasible in older adults with and without MCI. Sleep/wake rhythms variables generated from this system correlated with cognitive function, suggesting future studies can use this approach to evaluate novel, scalable, risk factor characterization and targeted therapy approaches.