Skip to main content

Table 2 Mean performance of the advanced models during cross-validation as well as performance during external validation. CV = Cross-validation

From: Converting between the International Prostate Symptom Score (IPSS) and the Expanded Prostate Cancer Index Composite (EPIC) urinary subscales: modeling and external validation

Advanced Models

Target variable

Input variable(s)

Architecture

Best mean absolute error during CV

Mean absolute error on external data

Min absolute error on external data

Max absolute error on external data

EPIC Urinary Summary

All IPSS questions

Linear regression

9.83

9.88

0.31

27.45

EPIC Urinary Summary

All IPSS questions

Support vector regression

9.59

11.07

1.45

31.57

EPIC Urinary Summary

All IPSS questions

K-nearest neighbors regression

9.32

9.29

0.47

23.64

EPIC Urinary Summary

All IPSS questions

XGBoost

9.84

10.38

0.19

32.86

EPIC Urinary Irritative/Obstructive

All IPSS questions

Linear regression

6.77

6.42

0.26

19.11

EPIC Urinary Irritative/Obstructive

All IPSS questions

Support vector regression

6.46

6.36

0.07

21.31

EPIC Urinary Irritative/Obstructive

All IPSS questions

K-nearest neighbors regression

6.56

7.32

0.45

23.21

EPIC Urinary Irritative/Obstructive

All IPSS questions

XGBoost

6.81

6.87

0.10

31.17

IPSS total

All EPIC Urinary subscale questions

Linear regression

2.65

3.79

0.36

13.71

IPSS total

All EPIC Urinary subscale questions

Support vector regression

2.63

3.89

0.26

14.17

IPSS total

All EPIC Urinary subscale questions

K-nearest neighbors regression

2.86

4.92

0.00

21.50

IPSS total

All EPIC Urinary questions

XGBoost

2.69

4.22

0.27

18.56

IPSS total

Relevant EPIC Urinary questions

Linear regression

2.68

3.91

0.06

13.73

IPSS total

Relevant EPIC Urinary questions

Support vector regression

2.67

4.27

0.37

15.37

IPSS total

Relevant EPIC Urinary questions

K-nearest neighbors regression

2.75

4.28

0.33

14.78

IPSS total

Relevant EPIC Urinary questions

XGBoost

2.65

3.96

0.24

14.53