Fig. 4

The performance of different classifiers. a Taylor diagram assessing the quality of classifier predictions against the reference in the training set. b Forest plot showing the results of bootstrapping. c–k Receiver operating characteristic curves of different classifiers (adaboost, elastic net regression, gradient boosting machine, K-nearest neighbor, logistics regression, naive bayesian regression, neural networks, support vector machine and xgboost, respectively) in the test set. Adaboost adaptive boosting, xgboost eXtreme gradient boosting, AUC area under curve, CI confidence interval