Abstract Introduction Sleep disturbances are common in menopausal women, yet repeated in-laboratory polysomnography (PSG) is often costly and inconvenient. Home sleep monitoring may provide valuable insights in clinical trials. Sleepiz One+, a contactless home sleep monitor using radar technology, measures vital parameters and bodily movements during sleep. We evaluated the agreement and accuracy between the reference standard PSG and Sleepiz One+ in assessing sleep/wake status in women with menopause-associated sleep disturbances from the NIRVANA trial. Methods Postmenopausal women with sleep disturbances underwent simultaneous Sleepiz and PSG recordings for six nights (two consecutive nights at baseline, week 4, and week 12). PSG was conducted in qualified sleep laboratories by trained staff and scored manually by two independent treatment-blinded scorers, following American Academy of Sleep Medicine guidelines. Sleepiz data were analyzed automatically using proprietary algorithms. Time synchronization between modalities was achieved using PSG’s “lights-off”/”lights-on” markers and respiratory signal alignment. The final dataset included 237,258 epochs (30-second intervals) from 66 individuals. modalities were compared using epoch-level and parameter-level analyses. For epoch-level analysis, sleep/wake status was compared one by one between devices to calculate sensitivity, specificity, overall accuracy, and prevalence-adjusted bias-adjusted kappa (PABAK), presented as mean (95% confidence intervals CIs) across subjects. Parameter-level comparisons (e.g., wakefulness after sleep onset WASO, total sleep time, sleep efficiency) were calculated per night by each method and compared between devices. Results Epoch-level comparison showed that Sleepiz achieved a mean (95% CI) sensitivity of 0.91 (0.89, 0.94) for detecting sleep, a specificity of 0.64 (0.60, 0.68) for detecting wake, and an overall accuracy of 0.86 (0.84, 0.88) versus PSG. PABAK was 0.72 (0.68, 0.76), indicating substantial agreement. Parameter-level comparisons showed that Sleepiz underestimated WASO by 4.8 minutes, latency to persistent sleep by 9.6 minutes, total sleep time by 7.1 minutes, number and duration of awakenings by 0.4, and 1.0 minutes, respectively. Sleep efficiency was overestimated by 2.1%. Conclusion Epoch-level and parameter-level comparisons demonstrated good agreement between Sleepiz and PSG, supporting Sleepiz as a contactless home sleep assessment tool in clinical trials. Support (if any) Sponsored by Bayer. Medical writing assistance (Highfield, Oxford, UK) funded by Bayer.
Krämer et al. (2026) studied this question.