Skip to main content

Table 10 Overall accuracy (%), missed detection (%), and false alarm (%) rates of features with statistically significant performance

From: Evaluation of linguistic and prosodic features for detection of Alzheimer’s disease in Turkish conversational speech

  

Accuracy

Missed detection

False alarm

ID

Features

SVM

Bayes

CTree

NN

SVM

Bayes

CTree

NN

SVM

Bayes

CTree

NN

2.3

Pronoun freq.

73.4

72.2

59.5

54.4

39.3

39.3

46.4

57.1

19.6

21.6

37.3

39.2

2.8

Pronoun-to-noun ratio

72.2

70.9

65.8

58.2

39.3

39.3

50.0

60.7

21.6

23.5

25.5

31.4

5.3

Silence ratio

83.5

81.0

79.8

78.5

35.7

35.7

28.6

25.0

5.9

9.8

15.7

19.6

5.4

Silence to utt. ratio

73.4

69.6

62.0

69.6

35.7

42.9

42.9

39.3

21.6

23.5

35.3

25.5

5.6

Avg. silence count

81.0

78.5

69.6

68.4

21.4

17.9

32.1

39.3

17.7

23.5

29.4

27.5

5.7

Silence rate

73.4

72.2

72.1

68.4

64.3

67.9

35.7

39.3

5.9

5.9

23.5

27.5

5.8

Cont. speech rate

78.5

76.0

68.4

62.0

28.6

25.0

25.0

39.3

17.7

23.5

35.3

37.4

5.9

Avg. cont. word count

76.0

82.3

76.0

70.9

28.6

32.1

28.6

35.7

21.6

9.8

21.6

25.5

7.1

Phoneme rate

69.6

79.8

60.8

57.0

32.1

46.4

46.4

50.0

29.4

5.9

35.3

39.2

7.2

Word rate

62.0

65.8

55.7

50.6

25.0

32.1

50.0

53.6

45.1

35.3

41.2

47.1

  1. Features with statistically significant performance in age-, education-, and gender-controlled experiments. Performance of the classifier with highest accuracy is shown in italics for each feature. Note that because the number of patients and control subjects are not equal, sum of average error, which is (missed detection+false alarm)/2, and accuracy is not 100%.