EMG-based pattern recognition approach in post stroke robot-aided rehabilitation: a feasibility study

Table 5 GROUP II: classification results and confusion matrices for TEST2

Patient id and classification rate	Actual classes	Predicted classes
		North	East	South	West
P1	North	40	0	0	40
35%	East	0	0	0	0
	South	60	100	100	60
	West	0	0	0	0
P2	North	20	0	0	0
35%	East	60	20	0	0
	South	20	80	100	100
	West	0	0	0	0
P3	North	80	100	80	100
25%	East	20	0	0	0
	South	0	0	20	0
	West	0	0	0	0
P4	North	20	0	0	0
30%	East	0	0	0	0
	South	80	100	100	100
	West	0	0	0	0
P5	North	40	0	0	0
45%	East	60	40	0	40
	South	0	60	100	60
	West	0	0	0	0
P6	North	20	0	0	0
35%	East	0	0	0	0
	South	80	100	100	100
	West	0	0	0	0
P7	North	0	0	20	0
35%	East	20	0	0	0
35%	South	80	80	80	40
	West	0	20	0	60

Elements on the left-right diagonal indicate the percentage of correct classification, while elements on the off-diagonals denote percentage of misclassification. SVM was trained with data from GROUP I and validated individually on each patient with the muscle dataset selected for healthy subjects (see Methods).

ISSN: 1743-0003