P051 Evaluating sleep stage concordance of CardioRespiratary Sleep Staging (CReSS) on level III portable sleep study via simultaneous deployment of polysomnograph and level III portable sleep study

Abstract

Abstract Background Polysomnograph (PSG) has traditionally been the gold-standard for evaluation of sleep disorders; however, it is complex and costly. Alternatives to PSG have been available for some time with aim to simplify the process. One challenge of home sleep testing has been the ability to stage sleep. The conventional type 2 device does provide limited electroencephalogram (EEG) but is restricted by quality and efficiency. Use of photoplethysmography (PPG) to stage sleep has been validated in 2017. There is limited research for evaluating sleep stage concordance using simultaneous deployment of laboratory EEG and PPG. Aims To evaluate a commercially available, proprietary algorithm (CReSS) which utilises PPG based sleep staging in comparison with laboratory PSG EEG for sleep stage concordance when data is acquired simultaneously and processed in parallel post acquisition Results Confusion matrix analysis demonstrates an accuracy of 80% across all staging. Upon further analysis, precision and recall for wake, light sleep, deep sleep and REM sleep was 82% and 78%, 77% and 90%, 96% and 48%, and 83% and 77% respectively. When classifying sleep/wake and REM/non-REM the accuracy was 90% and 95% respectively. Comparison of sleep distribution and AHI v REI established data was statistically significant in the percentage of light sleep (p=0.004) and deep sleep (p=0.005) along with the AHI vs. REI (p=0.013). Conclusion It was observed that sleep stage concordance aligned with literature values and that REI was usually underestimated compared to the AHI.