Additional file 6 of Development of a predictive model to distinguish prostate cancer from benign prostatic hyperplasia by integrating serum glycoproteomics and clinical variables

Additional file 6: Confusion matrixes, confidence intervals and ROC curves for the remaining four ML algorithms (KNN, SVM, Logistic regression, Decision Tree). The following sets of variables were considered: (i) proteomic + clinical (named “Multivariate Analysis”, blue), (ii) proteomic variables only (named “Peptides Analysis”, green), (iii) clinical variables only (named “Biological Samples”, red), (iv) PSA only (named “Univariate Analysis”, red).