Fig. 4.

MetS components and systemic inflammation improve the prediction of gingival bleeding, but not that of periodontitis. Panel A AUROC curves showing the predictive power of GBM, RF, and XBG to discriminate periodontitis using only sex, age, and smoking habits (left), or with the addition of MetS factors and CRP (right). Panel B AUROC curves showing the predictive power of GBM, RF, and XGB to discriminate gingival bleeding using only sex, age, and smoking habits (left), or with the addition of MetS factors and CRP (right). Panels C, D. Bar graph of the relative importance of individual variables in the learning process of algorithms to identify periodontitis (panel C) and gingival bleeding (panel D)