Optimization Based Feature Selection Algorithm with Twin-Bounded Support Vector Machine for Medical Dataset Classification

Abstract

The medical business demands innovative technical tools to objectively examine information.Currently, there is a lot less emphasis placed on the value of knowledge in medicine as a work needing computer assistance in everyday clinical scenarios.Now that the healthcare industry has expanded beyond the confines of the medical environment, clinical decision support systems are increasingly often used to refer to AI systems.The existing system has issue with optimized feature selection process and error rates lead poor classification performance. Also, the earlier stages of disease prediction are not achieved effectively. The Adaptive Firefly Optimization Algorithm (AFOA) and Twin Bounded Support Vector Machine (TBSVM) algorithm are therefore introduced in this study to address the aforementioned issues.The K-Means Clustering (KMC) technique is used to pre-process the datasets once they have been gathered. It is used to effectively manage the missing data and error rates.The AFOA method, which produces the best fitness values using an objective function, is then used to choose the features.The emphasis is on choosing the greatest characteristics through the best fitness values. The TBSVM technique is then utilized to classify the medical dataset.The inclusion of a regularization term to structural risk minimization, or TBSVM, aims to maximize the margin.Additionally, it helps to boost classification accuracy and save training time.It was found that the suggested AFOA-TBSVM algorithm delivers superior performance than the current algorithms in terms of the greater accuracy, sensitivity, and specificity as well as the shorter execution time. These findings were based on the results of the experiments.