Fig. 4

ROC (receiver operating characteristic) curves with an area under the ROC curve (AUC) for the RFM-L algorithm pyruvate, lactate, acetate, α-diversity (Shannon index), and butyrate, as excellent joint predictors/discriminators between pre- and post-intervention in HITB. FPR, false-positive rate; HITB, high-intensity training and use of probiotic cheese group; RFM-L, Random Forest machine-learning; TPR, true positive rate