Table 4.

Model performance in the training, internal and external validation sets for in-hospital mortality for patients over 50.

Test Sets	Statistics	Lasso LR	Elastic net LR	XGBoost
Training	Sensitivity (95% CI)	0.06 (0.02,0.14)	0.25 (0.16,0.36)	0.99 (0.93,1.0)
	Specificity (95% CI)	0.98 (0.96,0.99)	0.95 (0.91,0.97)	1.0 (0.99,1.0)
	AUC (95% CI)	0.68 (0.62,0.75)	0.7 (0.64,0.77)	1.0 (1.0,1.0)
Internal validation	Sensitivity (95% CI)	0.05 (0,0.3)	0.15 (0.03,0.38)	0.45 (0.23,0.68)
	Specificity (95% CI)	1.0 (0.95,1.0)	0.97 (0.9,1.0)	0.91 (0.82,0.97)
	AUC (95% CI)	0.66 (0.54,0.79)	0.73 (0.61,0.84)	0.72 (0.6,0.85)
External validation	Sensitivity (95% CI)	0.05 (0.01,0.08)	0.27 (0.2,0.34)	0.43 (0.35,0.51)
	Specificity (95% CI)	0.97 (0.93,1.0)	0.95 (0.9,0.99)	0.89 (0.82,0.95)
	AUC (95% CI)	0.8 (0.75,0.86)	0.82 (0.76,0.87)	0.71 (0.66,0.77)

The cutoff threshold to determine sensitivity and specificity was 0.5