Comparison of rest-activity rhythm metrics derived from Oura ring and actiwatch devices: A device validation and analytical pipeline comparison study.

Objectives

Non-parametric rest-activity rhythm (RAR) metrics are increasingly used as 24-hour behavioral markers of circadian and general health. Although traditionally derived from research-grade actigraphy, consumer wearables may provide scalable alternatives for long-term monitoring. This study evaluated the performance of RAR metrics derived from Oura Ring versus Actiwatch (AW2) during concurrent wear and examined the influence of analytic pipeline.

Methods

Forty-one healthy adults (mean age 26.3 years; 27 female) wore both devices for 14 days. Interdaily stability (IS), intradaily variability (IV), relative amplitude (RA), and L5/M10 timing were calculated using nparACT, MotionWare, and a custom pipeline. Performance was evaluated using ICC(A,1), ICC(C,1), Spearman's ρ, and Bland-Altman analysis. Oura MET was zero-referenced from its resting floor of 0.9 for RA analysis.

Results

Participants contributed a mean of 10.6 valid concurrent wear days. Oura Ring had slightly higher IS and lower IV than AW2. IS, IV, L5/M10 timing, and RA all showed moderate-to-excellent absolute agreement (ICC[A,1] = 0.67-0.98) and strong rank-order associations (Spearman's ρ = 0.71-0.98). Estimates were similar across the 3 analytical pipelines, although differences in M10/L5 implementation influenced timing estimates in one outlier.

Conclusion

Oura-derived RAR metrics showed acceptable overall agreement with AW2 when the known difference in measurement origin was accounted for in the RA calculation. Results were generally consistent across analytic pipelines, supporting analytic reproducibility. The M10 outlier showed that unusual activity values on individual days can affect M10 estimates across pipelines.