Figure 2.

Tuning of the machine learning algorithms to classify the 4-category survival outcome. We tuned the random forests (RF), support vector machine (SVM) with linear or radial basis function kernel, and multilayer perceptron (MLP) to classify the 4-category outcomes using 5-fold cross-validation (Heatmap graphs: A-C, RF models; D-F, MLP models; G-I, radial basis function SVM; and J-L, linear SVM models). The left column was the data produced using transcriptomic features alone, middle column using clinical features alone, and the right column using clinico-transcriptomic features (merged dichotomized clinical and transcriptomic features).