2015 | OriginalPaper | Chapter
Adjusting Covariates in CRIB Score Index Using ROC Regression Analysis
Authors : Maria Filipa Mourão, Ana C. Braga, Alexandra Almeida, Gabriela Mimoso, Pedro Nuno Oliveira
Published in: Computational Science and Its Applications -- ICCSA 2015
Publisher: Springer International Publishing
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
In medical studies, the receiver operating characteristic (ROC) curve is a tool of extensive use to analyze the discrimination capability of a diagnostic variable. In certain situations, the presence of related covariate, continuous or categorical, to the diagnostic variable can increase the discriminating power of the ROC curve [
3
].
The Clinical Risk Index for Babies (CRIB) scale, appeared in 1993 to predict the mortality of babies with very low birthweight (VLBW) and/or less than 32 weeks of gestation [
2
]. Braga and Oliveira [
1
] concluded that this index performs well in computing the risk of death for VLBW infants (
$$< 1500$$
g).
In previous works, the authors studied the effect of the baby’s sex [
17
] and the mother’s age [
18
] on CRIB scale, using results of an intensive care unit of a Portuguese hospital.
In the present work, we propose to analyze the discriminative power of CRIB scale, using ROC regression analysis with GLM (Generalized Linear Models), in the classification of babies with and without the presence of covariates (newborn gender and mothers age).
This study is carried out using a random sample obtained from data collected during the period from 2010
$$-$$
2012. The data source was the “Portuguese VLBW infants network” that encompasses all newborns with less than 1500 g or 32 weeks of gestational age born in Portugal.