Enhanced Imputation Method Combining Single and Multiple Methods to Handle Missing Values in Microarray Data

Abstract

Gene Expression Classification (GEC) is a modern healthcare approach for enhancing present medical practices by classifying patient’s gene structure to different types of cancer so as to provide effective and personalized treatments especially for all types of cancer. The GEC system aids medical practitioner in providing personalized treatments. The proposed GEC system assess the gene structure of a cancer patient through highly intensive computational intelligence technique named Genetic Algorithm (GA). In GA, the search space is composed of candidate solutions to the problem i.e. the collection of gene expression in the corpus, which is going to be used for training the computation model, which can further be used for testing new cancer patients in order to make accurate prediction about the presence of cancer cells. This will enable doctors to treat different cancer patients differently. In this proposed approach, each gene expression has been represented by a vector termed as chromosomes. In each generation, the chromosomes are selected randomly and fitness is evaluated. The probabilistic similarity function is used to estimate the fitness of the chromosome to predict the patient health condition. Experimental results show that the proposed approach works with relatively better accuracy compared to that of baseline approaches.