Fig. 1.

Feature importance for all variables in the preliminary validation dataset. The red line represents the accuracy of the model, in the preliminary validation set, when the model is derived (using all patients of the preliminary training dataset) excluding all the features with a smaller importance parameter than a given feature. No statistically significant degradation of the accuracy is observed until the first 8 most important features are used, as schematically indicated by the dashed line. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)