Fig. 5From: Antibody selection strategies and their impact in predicting clinical malaria based on multi-sera dataOptimal data dichotomization antibody selection results. A Sensitivity versus specificity plot for each antibody according to the cut−off that maximized the Pearson’s χ2 statistic. B Statistical significance of each antibody following p−value correction using the Benjamini−Yekutieli procedure. C AUCs for the individual models: Logistic regression (LRM), Random Forest (RF) and XGBoost (XGB) embedded in the Super Learner; and the overall AUC provided by the Super Learner. D Confusion matrix of the predicted versus observed individual’s classification derived from the Super Learner model using the ROC01 and E) SpEqualSe criterionBack to article page