Explainable Artificial Intelligence with Metaheuristic Feature Selection Technique for Biomedical Data Classification

Abstract

AbstractRecently, explainable artificial intelligence (XAI) becomes a hot research topic due to its decision-making abilities in several real-time applications, particularly health care. The application of XAI approaches can be used to investigate the biomedical data for effective disease diagnosis and classification. At the same time, the high dimensionality of the healthcare data poses a curse of dimensionality problem which can be solved by the design of optimization algorithms. Therefore, this paper introduces an XAI with feature selection technique for biomedical data classification (XAIMFS-BMC). The proposed XAIMFS-BMC technique intends to proficiently categorize the biomedical data into distinct classes. In addition, the XAIMFS-BMC technique involves the design of chaotic spider monkey optimization (CSMO) algorithm for effective selection of feature subsets. Moreover, the deep neural network (DNN) is exploited for medical data classification, and its efficiency can be further improved by the use of Nadam-optimizer-based hyperparameter tuning process. The performance validation of the XAIMFS-BMC technique is tested using distinct benchmark medical dataset, and the results are inspected under several aspects. The comparative results reported the supremacy of the XAIMFS-BMC technique over the other techniques in terms of different performance measures.KeywordsExplainable artificial intelligenceFeature selectionMetaheuristicsBiomedical data classificationDeep learning