Classification results on the validation sets and their corresponding ROC curves. Top illustrates scatter plots of the logistic regression classifier (first layer) outcomes on a validation subset for subtype (A) 7q (subset5), (B) NPM1mutant (subset2), and (C) FLT3ITD/NPM1wt (subset3). Red and blue solid lines indicate the classification boundaries of the first layer (GEP and DMP respectively). The purple solid line indicates the classification boundary of the second layer. Black encapsulated circles indicates samples that are misclassified when we trained the first layer on GEP or DMP only. These are however correctly classified when we incorporate the the second layer. Bottom illustrates the ROC curves of the scatter plots.