Fig. 2. Histogram of the ratio of the 90% upper Bayesian prediction bound to the scheduled operating room (OR) time for all cases. From the 65,661 cases, 25,000 random samples were taken with replacement. was applied. The histogram shows the 24,004 for which the 90% upper prediction bound was between 75% and 225% of the scheduled OR time. If the historic data were not used and only the scheduled OR time ( i.e., nk = 0 in 2,3,4), all cases would have a 90% upper prediction bound equal to 149% of the scheduled OR time. There is a peak with many cases at that percentage, because this best overall percentage is used for those cases from combinations of surgeons and procedure(s) with little or no historic data.