Efficient feature selection for breast cancer classification using soft computing approach: A novel clinical decision support system

One of the essential data pre-processing methods for enhancing the performance of machine learning (ML) models is feature selection. Because they choose the most optimal features for ML problems, metaheuristic feature selection algorithms have gained popularity recently. The Gravitational Search Optimization Algorithm (GSOA), Emperor Penguin Optimization (EPO), and an integrated (hGSEPO) algorithm that combines GSOA and EPO are three metaheuristic feature selection algorithms that are presented in this paper. GSOA performs the global search in hGSEPO, and Emperor Penguin Optimizer (EPO) performs a more thorough local search. In order to find influential features while eliminating irrelevant features and reducing complexity, this article introduces a pioneering hybrid approach that combines the two distinct algorithms GSOA and EPO. While the baseline algorithms have been employed for feature selection in a few ML tasks, the hybrid of these two has been used for the first time for breast cancer (BC) classification. The reason for selecting BC as a case of investigation is due to its recognition as the second leading cause of death in women. According to earlier research, the feature selection (FS) stage is crucial when processing large datasets with the goal of forecasting medical conditions like BC. Based on the selection of the most important features necessary to achieve enhanced accuracy, this intelligent classification system divides the data from the benchmark BC Wisconsin Diagnostic Breast Cancer (WDBC) feature set into two classes. Additionally, the intention of the research is to ascertain the minimum quantity of features necessary to attain a higher level of accuracy. The experimental results show that the proposed approach works auspiciously and categorizes with astounding results, with the highest accuracy of 97.66%, 0.9687 sensitivity, 1.000 specificity, 1.000 precision, 0.9516 F1-score, and 0.9980 area under the curve (AUC).