[BAS04] PLS-Cox model: Application to gene expression data

Conférence Internationale avec comité de lecture : Proceeding of the 16th symposium on Computational Statistics, January 2004, pp.655-662,

Auteurs: P. Bastien

Résumé: With advances in high-density DNA microarray technology, gene expression profiling is extensively used to discover new markers and new therapeutic targets. This technique supposes to take into account the expression of thousands of genes with respect to a limited number of patients. To predict survival probability on the basis of gene expression signatures can become a very useful diagnostic tool. In the context of highly multidimensional data the classical Cox model does not work. The PLS-Cox model by operating a dimension reduction of the gene expression space directed towards the explanation of the risk function appears particularly useful. It allows the determination of signatures of genomic expressions associated with survival, to predict the survival probability from these profiles, and reduce inter individual variability by changing the level of adjustment from a phenotypical level to a genotypical level.


@inproceedings {
title="{PLS-Cox model: Application to gene expression data}",
author=" P. Bastien ",
booktitle="{Proceeding of the 16th symposium on Computational Statistics}",