Skip to main content

Table 3 Evaluation of model performance in detecting “Pathology” chest radiographs

From: Classification of chest radiographs using general purpose cloud-based automated machine learning: pilot study

 

Pathology

Not pathology

Total

 Classified as “Pathology”

251 (true positive)

81 (false positive)

332

 Not classified as “Pathology”

81 (false negative)

224 (true negative)

305

 Total

332

305

 

Statistics

Value

95% confidence interval

 Sensitivity

75.60%

70.62 to 80.13%

 Specificity

73.44%

68.11 to 78.31%

 Positive likelihood ratio

2.85

2.34 to 3.46

 Negative likelihood ratio

0.33

0.27 to 0.41

 Positive predictive value

75.60%

71.80 to 79.04%

 Negative predictive value

73.44%

69.34 to 77.18%

 Accuracy

74.57%

71.00 to 77.91%