Figure 4.

Example Classification Tree based on ApSE analysis for a training set. For this particular gene cluster, key decisions were based on the values of ApSE at R = 0.1 (E1) and at R = 1.4 (E14). All subjects whose ApSE value at R = 1.4 was greater than 0.02 were classed as belonging to the cancer group. For subjects with an ApSE value at R = 1.4 less than 0.02 were further split into two groups. If the ApSE at R = 1.4 was above 0.01, they were assigned to the perilesional group. In the subset of subjects whose ApSE at R = 1.4 was less than 0.01, the ApSE at R = 0.1 was used to make the final decision: subjects whose ApSE at R = 0.1 fell below 0.33 were classed as belonging to the normal group, with the remainder (ApSE at R = 0.1 > 0.33) falling into the perilesional group. These decision rules were then applied to a corresponding validation set of subjects (for the same cluster of genes) to predict the biopsy class.