Identification of Microbial and Proteomic Biomarkers in Early Childhood Caries

Figure 7

Receiver operating characteristic (ROC) curves for three classification methods built for the CM-10 low-intensity proteomic data. ROC curves reflect the tradeoffs in between sensitivity and specificity for caries-active detection. A higher curve generally indicates a better method. The AUC (area under the curve) statistic summarizes the tradeoffs across varied sensitivity/specificity range. Standard deviations (sd) of the statistic are also reported. The random forest model appears to be the most effective classification method.