Table 3 Classification performance for linear SVM in each state between MDD and HCs and among MDD subgroups.

From: Identification of suicidality in patients with major depressive disorder via dynamic functional network connectivity signatures and machine learning

Groups

State

Acc

AUC

Sensitivity, %

Specificity, %

Cut off Point

Number

acc-P

MDD-HCs

State 1

80.36

0.12

0.18

0.64

0.14

1

0.994

 

State 2

81.33

0.72

0.64

0.76

0.49

9

0.023*

 

State 3

82.76

0.31

0.42

0.47

0.19

1

1

 

State 4

86.73

0.82

0.78

0.76

0.59

156

0.001*

 

State 5

80

0.74

0.61

0.83

0.51

15

0.02*

 

State 6

86.84

0.57

0.60

0.60

0.36

1

1

SA-NS

State 1

64

0.71

0.50

0.93

0.47

23

0.17

 

State 2

64.71

0.58

0.71

0.65

0.46

1

0.172

 

State 3

80

0.88

0.88

0.74

0.64

13

0.012*

 

State 4

79.63

0.81

0.83

0.75

0.63

99

0.008*

 

State 5

75

0.80

0.86

0.67

0.57

6

0.035*

 

State 6

72.41

0.75

0.71

0.77

0.55

1

0.018

SI-NS

State 1

62.86

0.58

0.60

0.73

0.44

6

0.189

 

State 2

63.41

0.66

0.71

0.59

0.42

3

0.19

 

State 3

75

0.78

0.78

0.74

0.58

2

0.023*

 

State 4

62.75

0.61

0.70

0.58

0.41

99

0.15

 

State 5

72.09

0.72

0.68

0.78

0.53

46

0.027*

 

State 6

50.70

0.07

0.17

0.2

0.04

1

0.56

SA-SI

State 1

66.67

0.48

0.4

0.85

0.34

202

0.16

 

State 2

63.41

0.61

0.88

0.42

0.37

4

0.194

 

State 3

77.36

0.63

0.63

0.78

0.49

39

0.014*

 

State 4

56.14

0.57

0.57

0.63

0.36

3

0.386

 

State 5

69.23

0.57

0.43

0.8

0.34

10

0.124

 

State 6

68.12

0.74

0.68

0.71

0.48

1

0.046*

  1. Number, predictive accuracy as a function of the number of connections used in the best classification process.
  2. Acc accuracy, acc-P the results of permutation test, MDD major depressive disorder, SA suicide attempter, SI suicide ideation, NS neither SA nor SI, HCs healthy controls.
  3. *P < 0.05.