To determine the optimum weighting coefficient for each sleep stage, NPSG data from five OSA and four normal subjects were randomly chosen. Tables 2 and 3 show the time intervals that the selected subjects spent in each sleep stage. Visual comparisons of these time intervals are shown in Figures 1 and 2. Based on the sleep durations in each sleep stage for these subjects, the weighting coefficient for each sleep stage was initially assigned as: Stage I = 1; Stage II = 2; Stage III = 5; Stage IV = 7; REM = 4 and Wake = 0. This selection, while somewhat arbitrary, aimed at giving additional weight to stage IV, because it is present in normal subjects and often absent in OSA patients. These weighting coefficients were further adjusted until the mean of the sleep indices for normal and OSA subjects were significantly different (p