Portable Brain Computer Interface Sleep Monitor Compared with Polysomnography in Macroscopic and Microscopic Sleep Structures

Accurate assessment of both macroscopic and microscopic sleep structure is critical for diagnosing sleep disorders and advancing personalized interventions. However, most existing portable sleep-monitoring devices, while validated at the macroscopic level, have not been benchmarked against PSG for microscopic sleep features. This study systematically evaluates a novel portable brain–computer interface (BCI) device, TH25, against PSG across multiple levels of sleep architecture, using an AASM-standard montage (F3, F4, E1, E2, A1, A2) with patch-based dry electrodes. Thirty-one adults underwent simultaneous overnight recordings using both PSG and TH25 systems. The system’s stable dry electrode design and full-night analysis—including natural artifacts—further highlight the high quality and reliability of the recorded signals under realistic conditions. At the macroscopic level, sleep staging performance of TH25 using an automatic sleep staging algorithm demonstrated high agreement with PSG (overall accuracy: 91.2%; Cohen’s kappa: 0.85). Key sleep metrics such as total sleep time, sleep efficiency, and stage distributions were closely aligned across devices. At the microscopic level, TH25 accurately detected sleep spindles and slow waves with comparable precision and F1-scores to PSG. These results demonstrate that the TH25 device provides a portable, cost-effective, and user-friendly solution for home-based sleep monitoring, while maintaining PSG-level accuracy.