Table 5 Performance evaluation metrics for each classification method across four severity classes of depression and anxiety.

From: Recognition of anxiety and depression using gait data recorded by the kinect sensor: a machine learning approach with data augmentation

  

TPR (recall) %

FNR (false neg rate) %

PPV (precision) %

FDR (false disc rate) %

F1 score %

Overall accuracy %

LDA

Mild

77.77

22.22

67.63

32.36

72.35

65.14

Minimal

42.22

57.77

50.33

49.66

45.92

Moderate

52.22

47.77

61.43

38.56

56.45

Severe

88.33

11.66

76.07

23.92

81.74

Naive Bayes

Mild

69.44

30.55

65.78

34.21

67.56

70

Minimal

62.77

37.22

70.18

29.81

66.27

Moderate

71.11

28.88

68.81

31.18

69.94

Severe

76.66

23.33

75.40

24.59

76.03

Multi-Class SVM

Mild

86.66

13.33

87.15

12.84

86.90

86.53

Minimal

85.55

14.44

84.15

15.84

84.84

Moderate

84.44

15.55

87.35

12.64

85.87

Severe

89.44

10.55

87.50

12.50

88.46

DNN

Mild

80

20

66.97

33.02

72.91

74.17

Minimal

51.11

48.88

70.22

29.77

59.16

Moderate

77.77

22.22

74.86

25.13

76.29

Severe

87.77

12.22

84.49

15.50

86.10

Back to article page