Fig. 3

Overview of model performance in terms of balanced accuracy in cross-validation (labeled as ‘CV’) and independent test sets (labeled as ‘Test’). Black dash line indicate random performance. Each category (Kidney and UC) consist of two independent clinical trial datasets. In each panel, the left end points indicate the model performance in CV trained on the indicated training set and the right endpoints indicate the performance in independent test set. A 5-fold cross validation was utilized in all experiments. The red line segments indicate the performance of our model GRRANN. Alternative models are group lasso (blue), ell ₁ regularized logistic regression (green), a multilayer perceptron (cyan) and a support vector machine (purple)