Identification of Microbial and Proteomic Biomarkers in Early Childhood Caries

Table 3

Performance statistics of three classification models tested on the combined microbial and MS proteomics data. The models were optimized for the average misclassification error (zero-one loss). For this experiment, only spectra for CM-10 low dataset were used and combined with the microbial data. The statistics include averages and standard deviations of test errors, sensitivities, and specificities of respective classifiers. The averages and standard deviations were calculated across 40 different train/test obtained through the random subsampling approach.

Classifier	Test error	Sensitivity	Specificity
“SVM”	8.91%±3.42%	89.61%±5.76%	92.36%±4.55%
SVM 100 WLCX	6.00%±2.67%	92.68%±4.46%	95.20%±3.87%
“RF”	16.05%±6.26%	76.52%±9.63%	91.14%±7.76%

SVM: linear support vector machine.
SVM on the top 20 Wilcoxon peaks.
Random forest.