The impact of actigraphy parameter choices on sleep estimates in one-month-old infants

INTRODUCTION: Actigraphy is a non-invasive method of recording sleep-wake cycles. Accurate detection depends on the choice of algorithm parameters, particularly the time window and activity threshold. This study aims to compare different combination of these parameters for estimating total sleep duration and the number of sleep bouts in one-month-old infants. METHODS: Forty-nine breastfed infants (aged 24.9 ± 3.6 days) were monitored at home for seven days. Sleep was recorded using an actigraph (MotionWatch8). Raw 60-s epochs were cross-checked with parental sleep report to correct artefactual immobility. Sleep-wake classification used 3- or 5-min windows and thresholds of 15, 20, or 40 counts. For each valid day, total sleep duration, number of sleep bouts, mean sleep bout duration and sleep-period durations were extracted. Analyses included linear mixed-effects models, Poisson regression, and plausibility screening based on pediatric literature values for sleep duration (9.3-20 h). A focused comparison between W3-T15 and W5-T15 configurations assessed sleep bout duration. RESULTS: Among the tested configurations, estimated sleep duration ranged from 15.9 to 18.1 h per day, with higher activity thresholds systematically inflating sleep duration. Threshold choice had a strong effect on total sleep duration (p < 0.001), whereas window length had minimal influence. Five-minute windows identified 50% fewer sleep bouts (p < 0.001), and produced substantially longer bout durations (p < 0.001). The W5-T15 configuration increased the proportion of sleep periods lasting 45-60-min sleep periods (7.8% vs. 5.2%, p < 0.001) and reduced the proportion of shorter periods (57.5% vs. 76.9%, p < 0.001). Overall, a threshold of 15 counts yielded the most coherent estimates. CONCLUSION: Actigraphy-derived sleep estimates in one-month-old infants are strongly influenced by the choice of activity threshold and time-window duration. Comparing six configurations revealed clear trade-offs: shorter windows increased sensitivity to brief transitions, whereas longer windows reduced fragmentation artefacts and yielded more consolidated sleep periods. These findings underscore the need for standardized parameter settings in early-infancy actigraphy and provide a reproducible basis for future methodological and validation work.