Analysis of Cancer Microarray Data using Constructive Neural Networks and Genetic Algorithms

The analysis of microarray data typically involves a feature selection method in order to select the most relevant genes while at the same time maximizing the information content. This work presents a methodology that use the Welch t-test to filter the number of initial features embedded in two different frameworks to select the predictor genetic profile: genetic algorithm and stepwise forward selection approaches. The genetic algorithm strategy combines mutual information and classification models to predict cancer outcome. Furthermore, a constructive neural network model, C-Mantec, is applied providing reduced network architectures with competitive results in comparison to other classifiers. Six free-public cancer databases are used to test our approach.