0951 Inter-Scorer Reliability Between Somno-Art Software and 5 Sleep Centers

Introduction A number of studies have evaluated epoch-by-epoch agreement between 2 technologists in scoring individual sleep stage. Scoring for the specific sleep stages average agreement ranged from 20% to 90% in different studies. The goal of this study is to evaluate the validity of the Somno-Art analyses against a pool of scorers including validated automated system for scoring polysomnograms. Methods Sixty polysomnography recording nights, representative of clinical practice (participants who are healthy or suffering from obstructive sleep apnea [OSA], or insomnia or major depressive disorder [MDD]), were scored by 5 different sleep scoring centers and by the Somno-Art Software (total of 6 scorers). Intra-class correlation coefficient (ICC) and Wilcoxon Signed-Rank Test were calculated between each scorer and the average value of the 6 scorers. In addition, epoch-by-epoch agreements between scorers were analyzed. Results Somno-Art Software estimation of sleep efficiency, wake, N1+N2, N3 and REM sleep belonged to the interscorer range for the full dataset and the participant groups, except N3 sleep in OSA patients. Somno-Art Software tended to overestimate sleep latency with a significant difference to the average scoring for insomniacs (4.7±1.6 min). On the full dataset, Somno-Art Software had good or excellent ICC scores for all sleep parameters except N3 sleep (moderate score). 4-stages epoch-by-epoch agreement ranged between 78.5-88.7% for the visual scorers and 69.9-71.2% for Somno-Art Software. For the healthy participants, visual scorers agreed between 68.9 and 86.9%, while Somno-Art Software overlapped this range with an interscorer agreement between 62.0 and 71.4%. Conclusion Somno-Art solution provides results that are comparable to manual and automatic scoring for commonly used metrics in sleep medicine. It shows a robust interscorer reliability in the range of the 5 polysomnography scorers Support (if any)