JMIR Publications

ABSTRACT

Background:

Digital phenotyping offers opportunities to capture real-time behavioral and physiological markers associated with mood disorders. For example, sleep and physical activity are two key behavioral exposures consistently shown to reduce depression and anxiety symptoms. However, feasibility and acceptability challenges, particularly relating to device burden and data reliability, remain barriers to upscaling sample sizes.

Objective:

This mixed-methods feasibility study evaluated the acceptability, usability, and data completeness of passive monitoring of sleep and physical activity via a smart ring and active ecological momentary assessment (EMA) of mood and contextual factors via a smartwatch in adults from the general population.

Methods:

Participants wore a consumer-grade Oura smart ring (Generation 3) to passively track sleep and physical activity for two weeks while concurrently completing brief mood EMAs (depression, anxiety and wellbeing) on a study-issued smartwatch. Feasibility metrics included device adherence, EMA completion rate, and patterns of missingness. Acceptability was assessed through semi-structured interviews. Exploratory analyses examined associations between daily mood, sleep and physical activity.

Results:

Adherence to both devices was high, with participants wearing the ring consistently (mean wear time = 97.3%, range = 91.8% to 98.9%) and completing most EMA prompts (mean compliance = 90%, range = 75.0%-100%). The smartwatch occasionally delivered a higher number of prompts than originally scheduled and showed some variability in timing of prompts. However, qualitative interviews indicated overall acceptability of both devices, with a slight preference for the smart ring, described as comfortable and unobtrusive. Some participants experienced the watch as bulky, and reported technical difficulties, overall indicating lower willingness to wear the smart watch again in future studies. There was no difference in the patterns of data missingness by time of day or day of the week. Mood ratings demonstrated reasonable variability, but exploratory associations between mood and either physical activity or sleep were not detected, likely due to limited statistical power.

Conclusions:

Passive monitoring via a smart ring and smartwatch-based EMA were both feasible, with high wear time and completion rates. Acceptability was higher for the smart ring, particularly in the context of concurrent device use. Smart rings appear well suited for scalable passive data collection, while future studies should ensure robust EMA scheduling mechanisms. Scaling this protocol up into larger samples is likely required to evaluate mood–behavior associations.