Abstract Introduction Manual actigraphy sleep scoring is time-intensive and requires managing multiple data sources simultaneously. Automated approaches are quicker but often lack validated, accurate nap detection and interpretability. Graphical user interface (GUI) applications such as ActiSleep Tracker and ActiLife address some issues but have performance and customizability limitations. We developed a novel open-source GUI application to streamline sleep scoring, implementing a user-friendly interface for a previously established manual scoring method. We preliminarily validated the application to establish sleep/wake metric and inter-rater reliability equivalence with the original method. Methods The application was developed to integrate data from wrist-worn Ametris wGT3X-BT devices, ActiLife epoch and non-wear sensor exports, the Sadeh sleep-wake algorithm, Choi non-wear algorithm, onset/offset scoring rules, and sleep diaries, all originally used for the manual method. Ten-day actigraphy data from thirty 3–5-year-old children in free-living conditions with minimal non-wear time were scored by three raters using both the manual (M) method and application (App). Scoring time was recorded using a separate application, and IDs were randomized. Inter-rater reliability for scorers’ onset/offset ranges and inter-method agreement for sleep-wake metric/algorithm implementations were evaluated with intraclass correlations assuming representative random sampling for raters and data (ICC (2, 1) ). Results The application improved scoring speed by 3. 3×, reducing mean scoring time by 14. 9 minutes/participant (n=88; M: 21. 5±8. 9 min; App: 6. 6±1. 8 min, p 0. 001, d=1. 72). For matched sleep periods including naps (n=391), both methods showed excellent interrater reliability ICCs (Appₒnset=0. 99, Appₒffset=0. 99; Mₒnset=0. 99, Mₒffset=0. 98). Excluding naps (n=295), manual was significantly less reliable for offset (Appₒnset=0. 95, Appₒffset= 0. 99; Mₒnset=0. 96, Mₒffset=0. 82. Inter-method agreement for sleep metrics from pooled, matched sleep periods from all raters (n=1240) yielded ICCs of 0. 99 (Total Sleep Time), 0. 90 (Wake After Sleep Onset), 0. 89 (Sleep Efficiency), 0. 98 (Time in Bed). However, all variance was explained by scorers’ different onset/offsets between methods (Lin’s CCC=1. 00), confirming algorithmic and metric equivalence. Conclusion The application demonstrates significantly faster scoring speed, maintains inter-rater reliability for naps and improves/maintains it for main sleep periods, while preserving algorithmic and metric equivalence. Its open-source nature and modular architecture support integration of numerous features. Future work will compare against consensus validation and incorporate additional algorithms. Support (if any) 1P01HD109876
Alam et al. (Fri,) studied this question.