Modeling of Explainable Artificial Intelligence with Correlation-Based Feature Selection Approach for Biomedical Data Analysis

Abstract

AbstractIn recent days, biomedical data analysis became a challenging problem due to the massive increase in the quantity of healthcare data. The emerging explainable artificial intelligence (XAI) tools can be applied for the effective examination of biomedical data and perform classification process. Besides, the high dimensionality of the medical data requires proper selection of features to reduce the complexity level. This paper presents an explainable artificial intelligence with correlation-based feature selection for biomedical data analysis (XAICFS-BDA) technique. The XAICFS-BDA technique aims to choose optimal features and then classifies data for biomedical decision making. In addition, the XAICFS-BDA technique performs the optimal selection of features using the correlation-based feature selection (CFS) technique. Besides, the classification of biomedical data is carried out by the use of fuzzy k-nearest neighbor classifier (FKNN), and the parameter tuning of this model is performed utilizing black widow optimization (BWO) approach. The experimental result analysis of the XAICFS-BDA technique is carried out using distinct benchmark biomedical dataset. Extensive comparative analysis pointed out the better performance of the XAICFS-BDA technique over the recent techniques.KeywordsBiomedical data analysisMedical data classificationXAIFeature selectionMachine learning