Table 2. Average model performances of five-fold cross-validation.

*Three features: platelet-lymphocyte ratio, eosinophils, and white blood cells.

PPV: positive predictive value; NPV: negative predictive value; ROC: receiver operating characteristics; AUC: area under ROC curves; PRAUC: area under the precision-recall curve.

Average model performances
	Scores with all features	Scores with three features*
Accuracy	0.726	0.744
Precision (PPV)	0.812	0.853
Recall (sensitivity)	0.777	0.760
Specificity	0.615	0.713
F1 score	0.791	0.797
NPV	0.595	0.612
AUC (predict)	0.696	0.736
AUC (predict_proba)	0.698	0.800
PRAUC	0.870	0.889