Table 3.

Although the AdaBoost algorithm achieved a competitive accuracy rate, its low MCC score indicated high false positive and negative.

Modality	Accuracy (mean ± STD)	MCC (mean ± STD)	AUC (mean ± S.D.)
Response-type-gaze+significant	71.0 ± 5.3	0.449 ± 0.118	0.758 ± 0.098
Response-type-gaze+significant (ANOVA)	57.6 ± 7.6	0.180 ± 0.102	0.659 ± 0.113

Accuracy rate, MCC, and AUC scores of the AdaBoost algorithm in recognizing ASD and ASD with ADHD symptoms in children. “ANOVA” indicates that the significant spatial features are determined by referring to p-value of ANOVA.