Table 2.

AUROC^a of machine learning models in predicting intravenous CIAKI^b.

Models	AUROC (95% CIc)	P valued
Logistic regression	0.690 (0.632-0.748)	.01
κ-Nearest neighbor	0.629 (0.566-0.693)	<.001
SVMe	0.644 (0.580-0.707)	<.001
DTf	0.633 (0.573-0.694)	<.001
RFg	0.726 (0.674-0.778)	.17
XGBh	0.665 (0.607-0.722)	.006
LGMi	0.651 (0.589-0.713)	<.001
RNNj	0.755 (0.708-0.802)	N/Ak

^aAUROC: area under the receiver operating characteristic curve.

^bCIAKI: contrast media–induced acute kidney injury.

^cCI: confidence interval.

^dCompared to the receiver operating characteristic curve of the RNN model.

^eSVM: support vector machine.

^fDT: decision tree.

^gRF: random forest.

^hXGB: extreme gradient boosting machine.

ⁱLGM: light gradient boosting machine.

^jRNN: recurrent neural network.

^kN/A: not available.