Effects of SDB on the non-normally distributed features extracted from SpO2 pattern characterization and PRV analysis.

Two-sample t-tests for unequal variances are applied to the log-transformed data to obtain the p-values (illustrated by p-value1). Additionally, Mann-Whitney U tests are also applied to the original data to account for the non-normal data distributions (p-value2). Features are described by their mean ± standard deviation for children with and without SDB and their mean difference. For abbreviations see table 2. For simplicity, only the mean (M) and standard deviation (S) of the significant time varying features are represented. Two variables, represented by ‡, were transformed using Box-Cox because they contained zero values.Effects of SDB on the non-normally distributed features extracted from SpO2 pattern characterization and PRV analysis.