Table 2.

Performance of machine learning algorithms for predicting the ARDS in the training cohort of TBI patients.

Classification Models	AUC (95% CI)	Accuracy	Sensitivity	Specificity	PPV	NPV	F1 Score
XGBoost	0.989 (0.983–0.995)	0.952	0.947	0.960	0.959	0.946	0.953
Light GBM	0.710 (0.669–0.752)	0.675	0.676	0.682	0.681	0.677	0.674
Random Forest	1.000	0.998	1.000	1.000	1.000	0.997	1.000
AdaBoost	0.815 (0.782–0.849)	0.736	0.724	0.752	0.742	0.736	0.731
CNB	0.618 (0.572–0.663)	0.592	0.694	0.495	0.574	0.624	0.626
SVM	0.509 (0.462–0.556)	0.538	0.253	0.822	0.629	0.534	0.305

XGBoost, extreme gradient boosting; Light GBM, light gradient boosting machine; AdaBoost, adaptive boosting; CNB, complement naïve Bayes; SVM, support vector machine; PPV, positive predictive value; NPV, negative predictive value.