Linking gene-expression experiments with survival-time data.

Abstract

We apply a model-based clustering approach to classify tumour tissues on the basis of microarray gene expression. The association between the clusters so formed and patient survival (recurrence) times is examined. The approach is illustrated using the lung cancer data set of Wigle et al. (2002). We show that the prognosis clustering is a powerful predictor of the outcome of disease, in addition to the stage of disease at presentation.