Improving gene selection in microarray data analysis using fuzzy patterns inside a CBR system

Abstract

In recent years, machine learning and data mining fields have found a successful application area in the field of DNA microarray technology. Gene expression profiles are composed of thousands of genes at the same time, representing complex relationships between them. One of the well-known constraints specifically related to microarray data is the large number of genes in comparison with the small number of available experiments or cases. In this context, the ability to identify an accurate gene selection strategy is crucial to reduce the generalization error (false positives) of state-of-the-art classification algorithms. This paper presents a reduction algorithm based on the notion of fuzzy gene expression, where similar (co-expressed) genes belonging to different patients are selected in order to construct a supervised prototype-based retrieval model. This technique is employed to implement the retrieval step in our new gene-CBR system. The proposed method is illustrated with the analysis of microarray data belonging to bone marrow cases from 43 adult patients with cancer plus a group of three cases corresponding to healthy persons. © Springer-Verlag Berlin Heidelberg 2005.