Skip to main content

Table 3 Averaged classification performance before applying PCA on initial features

From: EEG dynamical correlates of focal and diffuse causes of coma

Epoch length

Number of epochs in a trial

Accuracy (Hard)

Accuracy (Soft)

Specificity

Sensitivity

1 s

10

0.51

0.52

0.42

0.66

20

0.53

0.54

0.45

0.68

40

0.53

0.53

0.41

0.69

100

0.53

0.53

0.41

0.70

All

0.57

0.57

0.49

0.69

5 s

10

0.50

0.49

0.44

0.67

20

0.50

0.50

0.41

0.68

40

0.52

0.51

0.44

0.67

100

0.58

0.54

0.51

0.66

All

0.57

0.57

0.46

0.72

20 s

10

0.51

0.51

0.41

0.69

20

0.52

0.51

0.38

0.70

40

0.53

0.53

0.51

0.64

All

0.59

0.59

0.46

0.77

  1. Classification results are for different epoch length (1 s, 5 s, and 20s) and different number of epochs in a trial (10, 20, 40, 100, and all epochs). All the results are averaged classifier performance over 500 random training and testing sets. In each realization, the subjects in training and testing sets are different