Validation of a Wireless, Self-Applied Device for Sleep Recording in the Paediatric Population

Purpose: Although polysomnography (PSG) is the gold standard for sleep evaluation, its use in children and adolescents is limited by poor compliance and the time-consuming setup required. For the first time in a paediatric population, this study tested the Sleep Profiler (SP), a self-applied ambulatory PSG device previously validated in adults. Patients and Methods: Simultaneous standard PSG and SP recordings were acquired from hospitalized patients (n=14), aged 11 to 15 years, and independently scored by two certified sleep medicine experts, blinded to patient identity. Epoch-by-epoch agreement between manually scored PSG, manually scored SP and automatically scored SP records was assessed using confusion matrices and Cohen’s kappa coefficient (κ), with PSG as the reference. Sleep macrostructure measures were compared using Bland–Altman analysis. Results: Manual scoring of SP data (SP1) showed substantial agreement with standard PSG across all stages, particularly for REM (κ=0.83), wake (κ=0.83), and N3 (κ=0.82). By contrast, automatic scoring algorithms (AA1 and AA2) provided by the device performed markedly worse (κ always below 0.70). SP1 resulted particularly strong in stage discrimination (specificity over 96% for wake, N1, N3, REM, and 88% in N2), with a sensitivity above 84% in all stages, except N1, which was lower at 55%. Moreover, Bland–Altman plots showed minimal bias and good agreement for sleep macrostructure parameters. These findings were further supported by a robust interrater reliability between scorers (κ=0.82). Conclusion: The SP represents a reliable tool for sleep monitoring even in contexts where low compliance and time-consuming setups may compromise the quality or feasibility of standard recordings, such as in paediatric populations. While automatic scoring requires further refinement, and relatively small sample size should be considered when interpreting the generalizability of the findings, its robust detection especially of REM and N3 sleep make it suitable for both clinical and research applications.