Skip to main content

Table 3 C-index of the seven different variable selection methods and four predictive models. (More accurate models have a higher C-index. The joint two best indices are in bold)

From: Development and validation of a risk index to predict kidney graft survival: the kidney transplant risk index

No

Variable selection

Predictive models

Cox

RSF

SVM

DT

Combination 1

EO

0.67

0.66

0.58

0.60

Combination 2

PCA

0.65

0.60

0.65

0.55

Combination 3

EN

0.65

0.67

0.53

0.60

Combination 4

EO PCA

0.61

0.62

0.52

0.57

Combination 5

EO EN

0.66

0.61

0.61

0.57

Combination 6

EN PCA

0.64

0.65

0.56

0.61

Combination 7

EO EN PCA

0.64

0.63

0.62

0.60

  1. EO Expert opinion; PCA Principal component analysis; EN Elastic net; RSF Random Survival Forrest; SVM Support Vector Machine; DT Decision Tree
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us