Skip to main content

Table 1 Degradation in classification accuracy across sensor positions

From: The effect of accelerometer location on the classification of single-site forearm mechanomyograms

Participant

Accuracy C3 (%)

A1-A3

A2-A3

A4-A3

A5-A3

  

ΔC 1,3

ΔC 2,3

ΔC 4,3

ΔC 5,3

1

69.8 ± 4.3

7.5 ± 4

6.5 ± 4

6.8 ± 3

1.9 ± 3

2

71.0 ± 5.9

4.0 ± 4

6.9 ± 5

1.5 ± 5

2.0 ± 4

3

69.6 ± 4.5

5.7 ± 4

6.6 ± 4

7.7 ± 5

7.4 ± 5

4

74.6 ± 2.7

11.4 ± 4

4.7 ± 4

10.0 ±

7.5 ± 4

5

75.1 ± 4.2

11.8 ± 4

12.9 ± 4

1.7 ± 3

4.5 ± 4

6

69.1 ± 3.9

7.2 ± 5

5.7 ± 4

9.1 ± 5

1.5 ± 4

7

76.2 ± 4.3

14.9 ± 3

17.4 ± 3

4.6 ± 5

2.1 ± 4

8

71.2 ± 4.4

2.0 ± 3

0.8 ± 2

11.1 ± 5

3.6 ± 5

9

73.1 ± 3.0

5.9 ± 3

5.6 ± 3

7.2 ± 4

1.0 ± 3

10

76.4 ± 3.4

7.7 ± 4

1.0 ± 4

6.4 ± 3

0.9 ± 3

11

76.8 ± 3.0

1.4 ± 3

1.7 ± 2

4.9 ± 3

1.7 ± 3

12

69.7 ± 3.9

14.1 ± 5

12.2 ± 2

4.6 ± 3

4.9 ± 3

Avg

72.7 ± 0.6

7.8 ± 5

6.8 ± 6

6.3 ± 4

3.0 ± 4

  1. Values shown are mean ± standard deviation across all trials. Bold and italic values indicate that C i and C3 are significantly different (p < 0.05). Italic values indicate that the accuracy of the locally-trained classifier at this location was lower than that of A3, suggesting that this recording site yields signals that poorly reflect FCR activity.