Table 2.

Confidence interval results of machine learning methods in suicidality prediction

Cohort	Methods	95% Confidence interval of evaluation metrics
		Accuracy		Precision		Recall		F1 score		AUROC
Discovery	NB	0.692	0.798	0.751	0.849	0.692	0.798	0.707	0.811	0.760	0.856
	XGB	0.777	0.871	0.773	0.867	0.777	0.871	0.774	0.868	0.758	0.851
	RF	0.866	0.938	0.866	0.938	0.866	0.938	0.860	0.934	0.831	0.913
	SVC	0.670	0.780	0.742	0.842	0.670	0.780	0.688	0.796	0.737	0.837
	DNN	0.755	0.853	0.802	0.890	0.755	0.853	0.766	0.862	0.792	0.882
Test	NB	0.678	0.792	0.744	0.848	0.678	0.792	0.698	0.810	0.764	0.866
	XGB	0.745	0.849	0.728	0.835	0.745	0.849	0.734	0.840	0.704	0.814
	RF	0.730	0.838	0.705	0.816	0.730	0.838	0.712	0.822	0.691	0.803
	SVC	0.769	0.869	0.751	0.855	0.769	0.869	0.739	0.846	0.681	0.795
	DNN	0.779	0.877	0.763	0.865	0.779	0.877	0.757	0.859	0.856	0.936