Soal

Using CBR Systems for Leukemia Classification

Juan M. Corchado and Juan F. De Paz

Departamento de Informática y Automática, Universidad de Salamanca Plaza de la Merced s/n, 37008, Salamanca, España {corchado, fcofds}@usal.es

Abstract. The continuous advances in genomics, and specifically in the field of transcriptome, require novel computational solutions capable of dealing with great amounts of data. Each expression analysis needs different techniques to explore the data and extract knowledge which allow patients classification. This paper presents a hybrid systems based on Case-based reasoning (CBR) for automatic classification of leukemia patients from Exon array data. The system incorporates novel algorithms for data mining that allow to filter and classify. The system has been tested and the results obtained are presented in this paper. Keywords: Case-based Reasoning, dendogram, leukemia classification, data mining

1. Introduction

During recent years there have been great advances in the field of Biomedicine [1]. The incorporation of computational and artificial intelligence techniques to the field of medicine has yielded remarkable progress in predicting and detecting diseases [1]. One of the areas of medicine which is essential and requires the implementation of techniques that facilitate automatic data processing is genomics. Genomics deals with the study of genes, their documentation, their structure and how they interact [2]. We distinguish different fields of study within the genome. One is transcriptome, which deals with the study of ribonucleic acid (RNA), and can be studied through expression analysis [3]. This technique studies RNA chains thereby identifying the level of expression for each gene studied. It consists of hybridizing a sample for a patient and colouring the cellular material with a special dye. This offers different levels of luminescence that can be represented as a data array. Traditionally, methods and tools have been...