Development and Interpretation of a Clinicopathological-Based Model for the Identification of Microsatellite Instability in Colorectal Cancer

Table 3

Performance of different predictive models to identify dMMR. Accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of five predictive models in the test set.

Model	Sensitivity	Specificity	Pos Pred value	Neg Pred value	Accuracy	AUC-ROC
LR	0.7170	0.7595	0.6667	0.8000	0.7424	0.7835
RF	0.8679	0.6962	0.6571	0.8871	0.7652	0.8584
NB	0.7547	0.6329	0.5797	0.7937	0.6818	0.7424
SVM	0.7736	0.6709	0.6119	0.8154	0.7121	0.8174
XGBoost	0.7170	0.7468	0.6552	0.7973	0.7348	0.8055