ApSE as a function of R. ApSE is calculated at 55 values of R ranging from 0.1 to 5.5 (only the first 20 points are plotted here). This figure depicts the ApSE values in a training set, where red, blue and green curves correspond to data from cancer, perilesional and normal biopsy samples. For simplicity, only the mean curve for each group is shown here. For prediction purposes, a separate ApSE curve is computed for each subject. Because of the between subjects variation in the ApSE curves, selecting optimal decision criteria by eye based on data for all subjects is not possible. Selection of classification rules in training sets and application of the rules in validation sets were therefore carried out using classification trees.
Botting et al. BMC Bioinformatics 2009 10:66 doi:10.1186/1471-2105-10-66