Table 2

Discrimination and reclassification of prediction models for outcomes on test cohort

Outcome	Model	AUROC (95% CI)	p*	NRI (95% CI)	p†	AUPRC
TBI
	LR	0.770 (0.698 to 0.841)	NA	NA	NA	0.492
	XGB	0.809 (0.743 to 0.876)	0.04	0.689 (0.427 to 0.951)	<0.01	0.552
	SVM	0.776 (0.708 to 0.844)	0.77	0.339 (0.072 to 0.607)	0.01	0.479
	RF	0.800 (0.735 to 0.865)	0.13	0.308 (0.047 to 0.569)	0.02	0.532
	EN	0.799 (0.732 to 0.867)	0.06	0.698 (0.441 to 0.954)	<0.01	0.564
TBI-I
	LR	0.820 (0.751 to 0.890)	NA	NA	NA	0.551
	XGB	0.838 (0.775 to 0.901)	0.28	0.539 (0.258 to 0.821)	<0.01	0.554
	SVM	0.812 (0.748 to 0.875)	0.66	0.729 (0.464 to 0.994)	<0.01	0.469
	RF	0.836 (0.772 to 0.899)	0.38	0.333 (0.058 to 0.607)	0.02	0.552
	EN	0.844 (0.779 to 0.910)	0.15	1.093 (0.845 to 1.342)	<0.01	0.606
TBI-ND
	LR	0.767 (0.690 to 0.844)	NA	NA	NA	0.482
	XGB	0.800 (0.727 to 0.873)	0.07	0.605 (0.326 to 0.884)	<0.01	0.496
	SVM	0.778 (0.704 to 0.852)	0.56	0.285 (−0.001 to 0.572)	0.05	0.477
	RF	0.809 (0.739 to 0.880)	0.03	0.194 (−0.059 to 0.448)	0.13	0.535
	EN	0.811 (0.741 to 0.882)	0.02	0.768 (0.496 to 1.039)	<0.01	0.551
TBI-D
	LR	0.664 (0.490 to 0.838)	NA	NA	NA	0.138
	XGB	0.714 (0.512 to 0.917)	0.64	−0.026 (−0.605 to 0.553)	0.93	0.094
	SVM	0.814 (0.718 to 0.910)	0.09	0.209 (−0.325 to 0.742)	0.44	0.140
	RF	0.889 (0.801 to 0.976)	<0.01	−0.204 (−0.742 to 0.334)	0.46	0.196
	EN	0.871 (0.764 to 0.978)	0.01	0.119 (−0.415 to 0.654)	0.66	0.293

*Comparing the AUROC and the logistic regression model.
†Comparing the NRI and the logistic regression model.
AUPRC, area under precision-recall curve; AUROC, area under the receiver operating characteristic curve; EN, elastic net; LR, logistic regression analysis; NRI, net reclassification index; RF, random forest; SVM, support vector machine; TBI, traumatic brain injury; TBI-D, traumatic brain injury with death; TBI-I, traumatic brain injury with intracranial injury; TBI-ND, traumatic brain injury with non-discharge; XGB, extreme gradient boosting.