Abstract
This cross-sectional study was conducted from August 2021 to January 2023 in the psychiatric unit of a major university hospital in western Japan. The aim was to evaluate the performance of Nemuri SCAN (NSCAN), a non-wearable mat sensor developed in Japan for sleep monitoring and compare it to that of polysomnography (PSG) among psychiatric patients—a population not previously studied using this technology. The performance of NSCAN compared to that of PSG was lower than that reported in a preliminary study. To improve the performance of NSCAN, we developed a logistic regression model (proposed model) by incorporating data on 10 patient characteristics into the NSCAN decision algorithm, the Cole–Kripke equation (Cole model). The agreement, sensitivity, and specificity were 77.8% vs. 78.8%, 97.3% vs. 94.5%, and 28.2% vs. 38.9% for the Cole model and the proposed model, respectively. Notably, the proposed model demonstrated higher specificity, indicating improved performance in accurately identifying wakefulness. These findings highlight the importance of including patient characteristics in sleep monitoring algorithms and support the potential application of NSCAN in psychiatric care settings.