Skip to main content

Table 2 Model accuracy assessed by internal validation for 7 different models

From: Determine the therapeutic role of radiotherapy in administrative data: a data mining approach

Model

Classification variables

Number of rules

Accuracy (CI)

Confusion matrix

 

Model 1

Fraction-size

10

94.6% (94.4%, 94.7%)

C to P*:

2.2%

Disease site

P to C:

3.3%

Body-region

 

Time from the 1st treatment

Model 2

Fraction-size

13

94.5% (94.3%, 94.6%)

C to P:

1.9%

Body-region

P to C:

3.6%

Time from the 1st treatment

Model 3

Fraction-size

10

94.2% (94.0%, 94.3%)

C to P:

2.4%

Disease site

P to C:

3.4%

Time from the 1st treatment

Model 4

Fraction-size

8

94.1% (94.0%, 94.3%)

C to P:

2.0%

Time from the 1st treatment

P to C:

3.8%

Model 5

Disease site

18

92.0% (91.8%, 92.2%)

C to P:

3.0%

Body-region

P to C:

4.9%

Time from the 1st treatment

Model 6

Body-region

11

91.0% (90.8%, 91.2%)

C to P:

2.4%

Time from the 1st treatment

P to C:

6.6%

Model 7

Disease site

10

90.1% (89.9%, 90.3%)

C to P:

3.8%

Time from the 1st treatment

P to C:

6.2%

  1. *’C’ refers to curative, ‘P’ refers to palliative.