Skip to main content

Table 6 Evaluation results for each tag and in total, for different methods (rule, CRF, LSTM) and different evaluation datasets (MedNLP, dummy EHR, and pathology reports). M, d, and P respectively denote training data of MedNLP, dummy EHR, and Pathology reports; M + d denotes that training data consist of MedNLP+dummy EHR, all stands for all of these three datasets; other machine learning methods use the target evaluation dataset as its training data. In each cell, F1-score, precision, and recall are shown (in values multiplied by 100). The best scores for each tag type for each evaluation metric are presented in bold typeface. All evaluations were done by four-fold cross validations

From: De-identifying free text of Japanese electronic health records

Evaluation Results on MedNLP dataset

tag type

#of tags

scores

Rule

CRF

CRF

d

CRF

P

CRF

M + d

CRF

all

LSTM

LSTM

d

LSTM

P

LSTM

M + d

LSTM

all

total

490

F1

84.23

82.62

43.85

0.71

26.40

67.34

83.07

41.26

0.43

67.35

57.03

prec

78.90

85.63

46.20

2.50

21.51

66.54

81.33

41.07

0.48

66.98

57.94

recall

90.42

79.95

42.33

0.41

59.76

68.38

86.12

41.57

0.38

68.17

56.34

age

56

F1

93.43

71.12

30.00

0.00

32.55

53.04

95.83

71.11

0.00

84.72

87.50

prec

96.00

78.24

37.50

0.00

26.93

56.85

95.83

71.11

0.00

84.72

87.50

recall

91.16

65.47

28.13

0.00

46.05

50.00

95.83

71.11

0.00

84.72

87.50

hospital

75

F1

84.73

87.09

43.25

0.00

26.02

70.04

66.67

13.33

13.89

66.67

41.67

prec

80.75

93.52

66.67

0.00

20.55

91.67

75.00

11.11

10.67

70.83

45.83

recall

89.90

81.71

27.50

0.00

53.06

60.42

62.50

16.67

20.00

63.89

38.89

person

0

 

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

sex

4

F1

50.00

16.67

16.67

0.00

14.65

25.00

0.00

20.00

0.00

25.00

25.00

prec

50.00

25.00

12.50

0.00

8.68

25.00

0.00

20.00

0.00

25.00

25.00

recall

50.00

12.50

25.00

0.00

50.00

25.00

0.00

20.00

0.00

25.00

25.00

time

355

F1

50.00

16.67

47.43

0.98

14.65

70.57

96.14

67.22

42.98

89.78

82.67

prec

50.00

25.00

45.16

2.50

8.68

65.46

95.00

66.26

39.46

88.68

81.53

recall

50.00

12.50

50.19

0.61

50.00

76.50

97.41

68.30

47.94

91.00

82.67

Evaluation Results on Pathology Report dataset

tag type

#of tags

scores

Rule

CRF

CRF

M

CRF

d

CRF

M + d

CRF

all

LSTM

LSTM

M

LSTM

d

LSTM

M + d

LSTM

all

all

71

F1

13.97

74.26

0.00

0.62

1.45

57.63

81.67

0.00

0.00

1.45

81.25

prec

8.65

86.72

0.00

1.47

10.00

64.98

86.88

0.00

0.00

10.00

82.48

recall

43.33

65.16

0.00

0.39

0.78

54.06

78.84

0.00

0.00

0.78

80.15

age

0

 

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

hospital

31

F1

31.19

0.00

0.00

0.00

0.00

0.00

25.00

0.00

13.33

0.00

58.33

prec

26.47

0.00

0.00

0.00

0.00

0.00

25.00

0.00

13.33

0.00

58.33

recall

41.28

0.00

0.00

0.00

0.000

0.00

25.00

0.00

13.33

0.00

58.33

person

224

F1

0.00

91.08

0.00

0.00

6.25

71.31

95.19

0.00

0.00

0.00

95.83

prec

0.00

95.83

0.00

0.00

10.00

74.79

95.19

0.00

0.00

0.00

95.83

recall

0.00

87.21

0.00

0.00

4.55

69.63

95.19

0.00

0.00

0.00

95.83

sex

0

 

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

time

40

F1

9.25

10.57

0.00

2.00

0.00

18.82

25.00

3.81

0.00

6.25

19.44

prec

5.25

16.67

0.00

1.79

0.00

20.83

25.00

6.67

0.00

10.00

19.44

recall

43.09

9.09

0.00

2.27

0.00

19.32

25.00

2.67

0.00

4.55

19.44

Evaluation Results on Dummy EHR dataset

tag type

#of tags

scores

Rule

CRF

CRF

M

CRF

P

CRF

M + d

CRF

all

LSTM

LSTM

M

LSTM

P

LSTM

M + d

LSTM

all

total

3017

F1

43.74

66.97

44.01

19.67

67.13

65.79

63.99

20.33

1.60

69.82

68.19

prec

42.89

66.77

67.35

56.72

67.60

68.27

68.76

26.68

2.22

72.79

80.26

recall

44.75

67.34

33.28

12.34

66.69

63.63

60.20

17.03

1.25

67.24

60.04

age

39

F1

51.13

48.46

29.35

0.00

38.87

33.82

50.00

22.38

0.00

50.00

41.67

prec

51.97

65.25

28.85

0.00

41.56

35.72

50.00

19.05

0.00

50.00

45.83

recall

50.46

53.74

30.00

0.00

36.71

32.50

50.00

32.38

0.00

50.00

41.67

hospital

170

F1

15.98

47.85

33.19

0.00

48.62

35.73

22.22

35.79

0.00

40.00

43.33

prec

10.07

53.18

38.75

0.00

44.91

35.90

28.33

34.48

0.00

37.50

45.83

recall

39.06

43.73

29.42

0.00

53.60

37.81

29.17

37.33

0.00

43.75

41.67

person

135

F1

0.00

26.96

0.00

0.00

28.36

15.48

50.00

0.00

0.00

45.83

37.50

prec

0.00

26.79

0.00

0.00

29.91

19.64

50.00

0.00

0.00

45.83

37.50

recall

0.00

30.71

0.00

0.00

27.99

13.39

50.00

0.00

0.00

45.83

37.50

sex

16

F1

93.75

35.92

29.17

0.00

90.08

33.93

0.00

40.00

0.00

50.00

50.00

prec

100.0

44.27

50.00

0.00

95.83

50.00

0.00

40.00

0.00

50.00

50.00

recall

90.00

43.13

20.83

0.00

85.63

27.08

0.00

40.00

0.00

50.00

50.00

time

2657

F1

49.48

71.28

42.14

21.20

70.60

68.33

83.93

51.97

48.89

85.70

88.20

prec

51.81

71.44

64.94

59.35

71.24

70.94

84.82

52.59

48.89

86.51

89.24

recall

47.38

71.15

32.08

13.58

70.00

66.08

83.29

51.46

48.89

84.93

87.23