Feature optimization using teaching learning based optimization for breast disease diagnosis

Abstract

Disease diagnosis is a significant challenge in the field of medical science because most of the medical datasets contain irrelevant and redundant attributes which are not mandatory to obtain an accurate estimate of the disease diagnosis. In this work, we have used Teaching Learning Based Optimization (TLBO) algorithm for feature Optimization in automatic breast disease diagnosis. We have used a naive Bayes classifier for finding the fitness of individual and Multilayer Perceptron (MLP), J48, random forest, logistic regression algorithms for estimating the effectiveness of the proposed system. The results confirmed that the expected scheme produced higher accuracy on Wisconsin diagnosis breast cancer (WDBC) data set to classify the malignant and benign tumors. In short, the proposed TLBO variant presents an efficient technique to optimize the features for sustaining data-based decision making systems.