Opening the black box: interpretable machine learning for predictor finding of metabolic syndrome

BMC Endocrine Disorders

Table 1 Performance evaluation of MetS risk prediction in the test set

Classification model	Accuracy(%)	Sensitivity(%)	Specificity(%)	Youden index	AUROC (95%CI)
LR	92.3	64.5	97.0	0.615	0.807(0.800 ~ 0.815)^a
RF	99.5	96.9	100	0.969	0.984(0.982 ~ 0.987)^b
XGBoost	99.7	98.5	99.9	0.984	0.992(0.990 ~ 0.993)

^aindicates AUROC values of the XGBoost model compared with LR, Z = 30.986,P< 0.001
^bindicates AUROC values of the XGBoost model compared with RF, Z = 3.920,P< 0.001

ISSN: 1472-6823