Table 5.

Classification performance in the leave-one-language-out scenario.

TRAIN	TEST	Features	BACC	F1	SEN	SPE
Spiral drawing
CO+HU+US	CZ	Handcrafted	0.54	0.51	0.62	0.46
		CNN	0.45	0.41	0.48	0.42
CZ+HU+US	CO	Handcrafted	0.50	0.74	1.00	0.00
		CNN	0.63	0.62	0.54	0.71
CZ+CO+US	HU	Handcrafted	0.56	0.47	0.44	0.67
		CNN	0.71	0.67	0.67	0.75
CZ+CO+HU	US	Handcrafted	0.65	0.67	0.88	0.41
		CNN	0.38	0.32	0.33	0.42
Sentence writing
CO+HU+US	CZ	Handcrafted	0.63	0.68	0.78	0.48
		CNN	0.54	0.58	0.80	0.29
CZ+HU+US	CO	Handcrafted	0.59	0.30	0.18	1.00
		CNN	0.51	0.72	0.82	0.20
CZ+CO+US	HU	Handcrafted	0.67	0.64	0.59	0.75
		CNN	0.60	0.46	0.38	0.83
CZ+CO+HU	US	Handcrafted	0.71	0.67	0.59	0.83
		CNN	0.63	0.46	0.33	0.92

TRAIN, training dataset; TEST, test dataset; BACC, balanced accuracy; F1, F1 score; SEN, sensitivity; SPE, specificity.