Table 3 Effects of AI prediction between train and test groups among various diseases in Project 1 (%).

From: Prediction of skin disease using a new cytological taxonomy based on cytology and pathology with deep residual learning method

Ā 

Classes

1

2

3

4

5

6

7

8

9

10

11

Train group

Sensitivity and recall

96.00

57.14

89.47

96.97

89.74

94.59

75.00

100.00

100.00

99.57

100.00

Specificity

100.00

100.00

100.00

99.57

99.86

99.86

100.00

100.00

99.44

96.30

100.00

PPV and precision

100.00

100.00

100.00

95.52

97.22

97.22

100.00

100.00

90.48

97.66

100.00

NPV

99.72

99.60

99.73

99.71

99.45

99.72

99.60

100.00

100.00

99.31

100.00

F1

97.96

72.73

94.44

96.24

93.33

95.89

85.71

100.00

95.00

98.60

100.00

Test group

Sensitivity and recall

76.92

25.00

90.00

76.47

80.95

70.00

14.29

85.71

85.71

95.83

80.00

Specificity

97.59

100.00

99.74

99.18

99.47

99.74

100.00

100.00

98.15

84.38

99.49

PPV and precision

68.97

100.00

90.00

89.66

89.47

93.33

100.00

100.00

72.00

90.20

80.00

NPV

98.38

99.25

99.74

97.84

98.95

98.44

98.50

99.75

99.20

93.10

99.49

F1

72.73

40.00

90.00

82.54

85.00

80.00

25.00

92.31

78.26

92.93

80.00

  1. 1: Other congenital malformations of skin; 2: Gonococcal infection; 3. Other dermatitis; 4. Unspecified viral infection characterized by skin and mucous membrane lesions; 5. Pemphigus; 6. Congenital ichthyosis; 7. Herpes viral [herpes simplex] infections; 8. Scabies; 9. Other predominantly sexually transmitted diseases, not elsewhere classified; 10. Lichen planus; 11. Pediculosis and phthiriasis; PPV positive predictive value, NPV negativeĀ predictive value.