A Novel Feature Selection Method for Microarray Data Classification Based on Hidden Markov Model

Abstract

In this paper, a novel approach is introduced for integrating multiple feature selection criteria by using hidden Markov model (HMM). For this purpose, five feature selection ranking methods including Bhattacharyya distance, entropy, receiver operating characteristic curve, t-test, and Wilcoxon are used in the proposed topology of HMM. Here, we presented a strategy for constructing, learning and inferring the HMM for gene selection, which led to higher performance in cancer classification. In this experiment, three publicly available microarray datasets including diffuse large B-cell lymphoma, leukemia cancer and prostate were used for evaluation. Results demonstrated the higher performance of the proposed HMM-based gene selection against Markov chain rank aggregation and using individual feature selection criterion, where applied to general classifiers. In conclusion, the proposed approach is a powerful procedure for combining different feature selection methods, which can be used for more robust classification in real world applications.