An approach to improve the performance of subject-independent BCIs-based on motor imagery allocating subjects by gender

Table 3 Performance for the SI-BCIs designs in classifier test stage

Target Subject	Type of SI-BCI design	LEFTMI vs RIGHTMI	LEFTMI VS REST	RIGTHMI VS REST
Target Subject	Type of SI-BCI design	Mean (SD)	Mean (SD)	Mean (SD)
Male Subject	BCI-Men	53 (5)	60 (3)*	54 (3)
	BCI-Women	54 (5)	58 (2)	58 (4)*
	BCI-All	58 (2)	55 (2)	55 (7)
Female Subject	BCI-Men	56 (2)	76 (1)	71 (3)
	BCI-Women	58 (4)	79 (3)*	77 (4)*
	BCI-All	63 (6)	73 (4)	73 (5)
Male Patient	BCI-Men	65 (2)	63 (1)	59 (2)
	BCI-Women	62 (2)	67 (3)*	69 (2)*
	BCI-All	64 (1)	61 (4)	64 (5)
Female Patient	BCI-Men	63 (4)	73 (3)	83 (7)*
	BCI-Women	59 (5)	77 (2)*	76 (3)
	BCI-All	62 (2)	74 (4)	76 (2)

This table shows the %ACCs from 2 healthy subjects and 2 stroke patients as target users. The mean and standard deviation (in parenthesis) was obtained for the 10×4-fold cross-validation for each tested condition. The best %ACC for each subject in every condition was marked in BOLD. The asterisk indicates the best gender-specific BCI design that was statistically significant better (*p < 0.05) than BCI-All.

ISSN: 1475-925X