Skip to main content

Table 2 Prediction success measures for overlapping variables

From: Multivariate meta-analysis of proteomics data from human prostate and colon tumours

 

Sensitivity

Specificity

Geometric mean*

Variables

Average

Std

Average

Std

Average

Std

~40 Stable

1.00

0.00

0.88

0.12

0.93

0.06

17 Overlap

1.00

0.00

0.89

0.10

0.94

0.05

17 Random

0.94

0.09

0.79

0.14

0.86

0.10

  1. *geometric mean of sensitivity and specificity
  2. Average prediction success measures for variables selected using the missing value method "ten lowest". A PLS model with a five fold cross-validation (according to Figure 2) was built for three sets of variables: ~40 stable variables (one list for each cross-validation round), 17 overlapping variables and 17 variables randomly selected from the stable ones.
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us