BACKGROUND AND AIM: Manual polysomnography (PSG) scoring is time-consuming and subject to inter-scorer variability. Automated systems such as Somnolyzer24X7 (The Siesta Group Schlafanalyse GmbH, Vienna, Austria) have been developed to improve efficiency and standardization, but real-world validation data remain limited. This study aimed to evaluate the performance of Somnolyzer in a clinical PSG cohort by comparing automated outputs with manual scoring by two independent human scorers.

METHODS: This methodological comparison study included 42 full-night diagnostic PSG recordings. Each study was scored independently by two experienced human scorers according to AASM criteria, and by Somnolyzer using automated analysis. Twenty-two PSG parameters were extracted, including sleep architecture measures, respiratory indices, and arousal indices. Agreement was assessed using intraclass correlation coefficients (ICC), and differences in PSG parameters across scoring methods were analyzed using the Friedman test with post-hoc comparisons.

RESULTS: Measures of sleep continuity and REM-related parameters did not differ significantly between scoring approaches (all p>0.05). Significant differences were observed for N1 %, N2 %, and N3 % and for multiple respiratory indices (all p<0.05). Somnolyzer showed excellent agreement with the human consensus for key respiratory parameters, including obstructive apnea index (ICC 0.97), apnea index (ICC 0.96), and apnea–hypopnea index (ICC 0.93). Agreement was also high for total sleep time (ICC 0.94), sleep efficiency (ICC 0.95), and REM percentage (ICC 0.90). Moderate agreement was observed for sleep stage proportions, whereas lower agreement was observed for the mixed apnea index and the arousal index.

CONCLUSION: Somnolyzer provides automated PSG scoring results that are largely consistent with expert manual scoring and may serve as a reliable first-pass analysis tool in clinical practice.