Adaptive local binary descriptor statistics: A novel feature set for heart disease diagnosis

Abstract

Summary Heart disease is among the most common causes of death worldwide, and it is critical to recognize the condition early on. Clinical professionals face a difficult task in determining whether or not a person is at risk of coronary artery disease since it necessitates extensive knowledge and the use of rigorous clinical tests. The goal of this study is to create a revolutionary heart disease prediction paradigm by going through three primary phases: “feature extraction, optimum feature selection, and prediction.” Initially, attributes such as “statistical and higher‐order statistical features” are retrieved from the provided input data. “local binary pattern (LBP), arithmetic mean based LBP, harmonic mean based LBP, and geometric mean based LBP features” are retrieved from the suggested higher‐order statistical characteristics. In addition, the original characteristics of the input datasets are retrieved. However, in this case, the “curse of dimensionality” appears to be the most serious problem, necessitating the extraction of the best features. As a result, a new lion using the suggested cub growth function algorithm selects the optimum features from the retrieved data. Finally, an optimized deep convolution neural network is used to forecast the specified characteristics.