A Novel Radiomics-Based Machine Learning Framework for Prediction of Acute Kidney Injury-Related Delirium in Patients Who Underwent Cardiovascular Surgery

Table 4

5-fold cross-validation results of AUC after multiple logistic regression (MLR).

Signature	AUC	Precision	Recall	-score	Accuracy	Delong test for AUC values
Training set
Fold-1	0.633	0.696	0.320	0.438	0.773	,
Fold-2	0.594	0.619	0.250	0.356	0.739
Fold-3	0.635	0.667	0.340	0.450	0.757
Fold-4	0.640	0.708	0.333	0.453	0.773
Fold-5	0.662	0.639	0.426	0.511	0.757
Average	0.633	0.666	0.334	0.442	0.760
Testing set
Fold-1	0.596	0.200	0.500	0.286	0.667	,
Fold-2	0.817	0.385	0.833	0.526	0.804
Fold-3	0.745	0.583	0.636	0.609	0.800
Fold-4	0.609	0.214	0.500	0.300	0.689
Fold-5	0.503	0.182	0.250	0.211	0.667
Average	0.654	0.313	0.544	0.386	0.725
Best model in validation set	AUC	Precision	Recall	-score	Accuracy
	0.753	0.667	0.277	0.391	0.934