Skip to main content

Table 12 Accuracy, missed detection, and false alarm rates of the best performing features

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

ID

 

Accuracy

Missed detection

False alarm

     

1 feature

      

5.3

SVM

83.5%

35.7%

5.9%

     
  

(73.5 to 90.9)

(18.6 to 55.9)

(1.2 to 16.2)

     

7.1

Bayes

79.8%

46.4%

5.9%

     
  

(69.2 to 88.0)

(27.5 to 66.1)

(1.2 to 16.2)

     

2.3

SVM

73.4%

39.3%

19.6%

     
  

(62.3 to 82.7)

(21.5 to 59.4)

(9.8 to 33.1)

     

2 features

      

5.3 and 7.1

SVM

83.5%

35.7%

5.9%

     
  

(73.5 to 91.0)

(18.6 to 55.9)

(1.2 to 16.2)

     

2.3 and 5.3

Bayes

82.3%

32.1%

9.8 %

     
  

(72.1 to 90.0)

(15.9 to 52.4)

(3.3 to 21.4)

     

2.3 and 7.1

Bayes

78.5%

39.3%

11.8%

     
  

(67.8 to 87.0)

(21.5 to 59.4)

(4.4 to 23.9)

     
  1. Performance of best performing single feature in each feature category, as well as performance of combinations of these features. Results are reported only for the best classifier. 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%. Confidence intervals are shown in parenthesis.