TABLE 3.

AUCs for the performance of the proteomics-based predictive models.

Cohort	Classifiers	AUC	Accuracy	Sensitivity	Specificity	PPV	NPV	F1 score
Training	LR	0.678 (0.585–0.771)	0.636	0.765	0.565	0.494	0.812	0.600
	RF	0.806 (0.735–0.877)	0.727	0.863	0.652	0.579	0.896	0.693
	XGBoost	0.696 (0.606–0.786)	0.650	0.784	0.576	0.506	0.828	0.615
	SVM	0.707 (0.613–0.800)	0.678	0.745	0.641	0.535	0.819	0.623
	MLP	0.760 (0.671–0.848)	0.783	0.529	0.924	0.794	0.780	0.635
	KNN	0.781 (0.705–0.857)	0.699	0.843	0.620	0.551	0.877	0.667
	DT	0.671 (0.590–0.753)	0.678	0.647	0.696	0.541	0.780	0.589
	LightGBM	0.751 (0.671–0.831)	0.699	0.745	0.674	0.559	0.827	0.639
Test	LR	0.577 (0.433–0.721)	0.532	0.560	0.514	0.438	0.633	0.491
	RF	0.555 (0.408–0.702)	0.581	0.560	0.595	0.483	0.667	0.519
	XGBoost	0.629 (0.489–0.768)	0.565	0.640	0.514	0.471	0.679	0.542
	SVM	0.615 (0.475–0.756)	0.581	0.600	0.568	0.484	0.677	0.536
	MLP	0.609 (0.467–0.751)	0.565	0.120	0.865	0.375	0.593	0.182
	KNN	0.515 (0.365–0.664)	0.597	0.600	0.595	0.500	0.688	0.545
	DT	0.549 (0.414–0.662)	0.565	0.400	0.676	0.455	0.625	0.426
	LightGBM	0.526 (0.368–0.684)	0.565	0.480	0.622	0.462	0.639	0.471

SVM: support vector machine, LR: logistic regression, RF: random forest, KNN: k-nearest neighbor, AUC: area under the curve, NPV: negative predictive value, PPV: positive predictive value.