Statistical Analysis for Visualization of Sleep Quality using Sleep index

Abstract

Many people complain of sleep problems in Japan. Therefore, many studies challenge to develop technologies for assessing and improving sleep quality. However, little studies have statistically described and visualized the major components underlying normal sleep structures based on physiological data. In this study, a laboratory-based polysomnography (PSG) was performed for consecutive two nights in 56 healthy young adults (male = 31, female = 25). To physiologically characterize sleep, the data of the second night were scored according to the standard rules. Sleep stages were classified into REM sleep and NREM sleep, the latter of which were further classified into light and deep sleep. Then the indexes such as the latency and percentage of these sleep stages were calculated and the correlations among these indexes were assessed. After principal component (PC) analysis was adopted for these indexes, three components were identified for describing normal sleep structure: a cumulative contribution ratio from the first PC to the third PC was more than 90%.