Table 2.

Model performance for the best-performed algorithm

Prediction model	Best algorithm	Model performance
		Accuracy	AUROC (95% CI)	Sensitivity	Specificity	PPV	NPV	F1	PLR	NLR	Scaled brier
Model 1	XGBoost	0.75	0.90 (0.89, 0.91)	0.899	0.657	0.621	0.913	0.74	2.63	0.15	0.297
Model 2	CatBoost	0.915	0.96 (0.95, 0.97)	0.918	0.913	0.754	0.975	0.83	10.63	0.09	0.609
Model 3	CatBoost	0.86	0.946 (0.94, 0.95)	0.886	0.857	0.662	0.959	0.76	6.22	0.13	0.462
Model 4	ET	0.95	0.987 (0.98, 1)	0.965	0.946	0.944	0.967	0.95	17.96	0.04	0.754
Model 5	ET	0.92	0.973 (0.95, 1)	0.881	0.943	0.881	0.942	0.88	15.36	0.13	0.696
Model 6	ET	0.95	0.993 (0.99, 1)	0.945	0.952	0.896	0.976	0.92	19.85	0.06	0.75