Table 1 Model performance metrics in the full sample (n = 160).

From: Classifying attentional vulnerability to total sleep deprivation using baseline features of Psychomotor Vigilance Test performance

Metric

Time since wake

4 h

6 h

8 h

10 h

12 h

14 h

Accuracy (%)

70.4 ± 1.7

66.7 ± 1.7

75.6 ± 1.9

72.3 ± 1.6

69.9 ± 2.1

64.8 ± 1.4

Cohen’s kappa

0.51 ± 0.03

0.50 ± 0.03

0.61 ± 0.03

0.54 ± 0.03

0.55 ± 0.03

0.48 ± 0.02

Classifying resilient participants versus bottom 75% of performers

Sensitivity (%)

62.9 ± 3.5

84.0 ± 3.0

81.0 ± 2.7

69.2 ± 2.4

83.7 ± 4.1

89.7 ± 2.7

Specificity (%)

91.0 ± 1.3

84.9 ± 1.2

90.7 ± 0.92

92.8 ± 1.2

81.4 ± 1.6

75.1 ± 1.2

Positive predictive value (%)

71.8 ± 3.6

66.9 ± 2.4

76.0 ± 2.2

77.6 ± 2.9

60.7 ± 2.7

55.0 ± 1.9

Negative predictive value (%)

88.5 ± 1.0

94.2 ± 1.0

93.7 ± 0.83

90.3 ± 0.76

94.1 ± 1.4

95.9 ± 1.0

Classifying vulnerable participants versus top 75% of performers

Sensitivity (%)

57.0 ± 3.4

73.9 ± 2.4

70.0 ± 3.5

57.4 ± 3.3

84.0 ± 4.2

79.6 ± 2.1

Specificity (%)

91.8 ± 1.2

83.2 ± 1.4

91.2 ± 1.7

94.4 ± 1.2

86.7 ± 1.5

85.4 ± 1.2

Positive predictive value (%)

71.4 ± 3.7

60.3 ± 2.6

74.6 ± 4.1

79.1 ± 5.0

69.3 ± 3.3

66.0 ± 2.5

Negative predictive value (%)

86.9 ± 0.94

90.9 ± 0.85

90.3 ± 1.1

87.2 ± 0.86

94.4 ± 1.4

92.8 ± 0.74

  1. A 3-class linear discriminant model was developed using features of Psychomotor Vigilance Test performance at each baseline time point to classify participants in different vulnerability groups (vulnerable, intermediate, resilient). For each model performance metric, the mean ± SD is shown for 100 runs of stratified 5-fold cross validation.
Back to article page