Fig. 3.

Results on the training part of Hatzis data. IBS values (a) for the clinical and the combined models and the differences (b) between the models as evaluated on the training data set. C-index values (c) for the clinical and the combined models and the differences (d) between the models as evaluated on the training data set. The gray squares indicate the values of the models developed on the whole training data set, whereas the boxplots indicate the values obtained from the subsampling analysis