Distribution of AUC values for gene ILV1-44 produced by the revolving sliding window method compared to all genes in the TNB data set. A set of 30 windows of length 20 were produced by the revolving window method for the first 30 and last 30 cases in the breast cancer data set, representing earliest events versus longest recurrence-free survival, respectively. Each of the 900 possible combinations were analysed to produce an accuracy score (AUC). Shown is the distribution of scores for one of the most predictive genes, ILV1-44, in comparison to the distribution curve for all other genes. The plot is representative of the AUC analysis for the other immune genes found to be predictive in the TNB study.