248 Behavioural Biometrics: Using Smartphone Keyboard Activity as a Proxy for Rest-Activity Patterns

Abstract

Introduction Rest-activity patterns are important aspects of healthy sleep and may be disturbed in conditions like circadian rhythm disorders, insomnia, insufficient sleep syndrome, and neurological disorders. Long-term monitoring of rest-activity patterns is typically performed with diaries or actigraphy. Here, we propose a fully unobtrusive method to obtain rest-activity patterns using smartphone keyboard activity. This study investigated whether keyboard activities from habitual smartphone use are reliable estimates of rest and activity timing compared to daily self-reports within healthy participants. Methods First-year students (n = 51) used a custom smartphone keyboard to passively and objectively measure smartphone use behaviours, and filled out the Consensus Sleep Diary for one week. The time of the last keyboard activity before a nightly absence of keystrokes, and the time of the first keyboard activity following this period were used as markers. Results Results revealed high correlations between these markers and user-reported onset and offset of resting period (r ranged 0.74 - 0.80). Linear mixed models could estimate onset and offset of resting periods with reasonable accuracy (R2 ranged 0.60 - 0.66). This indicates that smartphone keyboard activity can be used to estimate rest-activity patterns. In addition, effects of chronotype and type of day were investigated. Conclusion Implementing this monitoring method in longitudinal studies would allow for long-term monitoring of (disturbances to) rest-activity patterns, without user burden or additional costly devices. It could be particularly useful in studies amongst clinical populations with sleep-related problems, or in populations for whom disturbances in rest-activity patterns are secondary complaints, such as neurological disorders. Support (if any):

Smartphone Keyboard Interaction Monitoring as an Unobtrusive Method to Approximate Rest-Activity Patterns: Experience Sampling Study Investigating Interindividual and Metric-Specific Variations

“…In fact, the metric-specific variations in the deviations suggested smaller discrepancies between the bed-related versus the sleep-related estimates. Similar to the findings of Druijff-van de Woestijne et al [30], the keyboard-derived estimates of the timing of the onset of the rest and active periods and the duration of the rest period were, on average, closer to the bed-related estimates (self-reported bedtime, out-of-bed time, and total bed period) than to the sleep-related estimates (self-reported try-to-sleep time, sleep onset, sleep offset, and total sleep period). For all comparisons, except the comparison between the timing of the last keystroke and self-reported bedtime, the null hypothesis that the values were similar was rejected.…”

“…This study replicated earlier research findings [29][30][31], showing that the timing of the last keystroke and the first keystroke surrounding the nocturnal prolonged keyboard inactivity period on the smartphone can serve as good predictors of the self-reported timing of the rest and active period onsets. These findings also complemented the associations between smartphone interactions and putative sleep-wake timing derived from actimetry [29,31].…”

“…In fact, the findings suggested that approximately 75% of the variance in the total bed or sleep period remained unexplained. Earlier studies also revealed lower explained variances in the putative duration of the rest period, with similar percentages [29,30]. A potential explanation for these weaker associations is that the duration-related estimates are based on 2 events surrounding the prolonged inactive period, which are both likely prone to inaccuracies in the approximation.…”

“…In this study, in contrast to the relatively large discrepancy between KAP and the self-reported total sleep period, the equivalence tests revealed that equivalence was not rejected with a margin of +30 or −30 minutes for most of the other comparisons (except for the difference between the timing of the last keystroke and self-reported try-to-sleep time and sleep onset). Earlier studies also reported average discrepancies of <30 minutes between smartphone-based estimates and self-reported or actimetry-based proxies of sleep-wake timing [29][30][31]56]. This suggests that the approximations of rest-activity timing rendered by smartphone interaction monitoring are very similar to those rendered by commonly used methods among free-living persons in naturalistic settings when inspecting aggregated, mean-level data.…”

“…The widespread use of smartphones offers new opportunities to detect and predict behavioral manifestations related to rest-activity patterns in everyday situations in a noninvasive manner and on a large scale. Recent studies have shown that smartphone-derived metrics can serve as proxies for the timing of sleep and wake by detecting periods of activity and inactivity during the 24-hour day [29][30][31]. These studies reported strong correlations between keyboard or touchscreen interactions and estimates of rest-activity timing based on diaries or actigraphy but also revealed informative deviations between these different assessment methods.…”