Artificial neural network application for identifying risk of depression in high school students: a cross-sectional study

Table 2 Comparison of predictive performance of ANN, CHAID, Quest, C5.0, C&T Tree, Logistic regression, and Bayesian Network of testing set (N = 301)

	ANN	Logistic regression	Quest	C&T Tree	C5.0	CHAID	Bayesian Network
Accuracy (%)	81.06	80.07	78.41	77.74	76.41	73.42	70.43
Sensitivity (%)	65.3	64.2	46.3	62.1	54.7	64.2	55.7
Specificity (%)	88.4	87.4	93.2	85.0	86.4	77.7	80.7
Area under ROC curve (AUC)	0.846	0.843	0.719	0.783	0.773	0.806	0.711

Note: ANN, artificial neural network. CHAID, Quest, C5.0 and C&T Tree are decision tree algorithms. C&R Tree, Classification and Regression tree, CHAID, Chi-Square Automatic Interaction Detection

ISSN: 1471-244X