Skip to main content

Table 6 Breakdown of classification errors according to different error categories

From: Automatic classification of diseases from free-text death certificates for real-time surveillance

Category

Total #errors

% of total

#coroner

  

records

cases

Classification errors:

405

58.0

76

Word variations

75

11.5

12

Word combinations

98

13.0

15

Secondary causes

150

22.4

43

Class confusion

82

11.1

6

Ground truth issues:

290

42.0

130

Ground truth class confusion

98

12.7

10

Ground truth error

167

25.5

113

Ground truth empty

25

3.8

7

  1. Categories are divided into actual classification errors and other issues related to the use of ICD-10 codes as the ground truth label