Renyi Entropy Predictive Data Mining And Weighted Xavier Deep Neural Classifier For Heart Disease Prediction

Abstract

During the past few years, Frequent Pattern Mining (FPM) has received the interest of several researchers that necessitate extracting items from transactions, and sequences from datasets, clarifying heart disease diagnosis that materializes commonly, and recognizing specific arrangements. In this era with healthcare involving significant evolutions, the unforeseeable movement and enormous amount of data concerning the classification of heart disease lead the way to new issues in FPM, such as space and time complexity. However, most of the research work concentrates on identifying the healthcare patterns relating to heart disease that transpires frequently, where the patterns within every transaction were known a priori. To address such issues in the present scenario, selecting the predominant patterns or frequent patterns is essential using relevant FPM models. The primary objective of this work is to enhance FPM mining results and reduce the misclassification rate of Cardiovascular Disease (CVD) dataset samples. This work proposes a novel method called Renyi Entropy Homogenized Weighted Xavier-based Deep Neural Classifier (REHWX-DNC) for heart disease prediction. To tackle the first challenge, the Renyi Entropy-based Frequent Pattern Mining (RE-FPM) algorithm is proposed, which filters the low-quality features using the Renyi Entropy function. To handle the second issue, the HWX-DNC model is designed to assist in minimizing the misclassification rate by employing the Swish activation function. A dataset for CVD synthesis can be analyzed to obtain significant accuracy for this study, and REGEX-DNC can be improved with compared state-of-the-art methods. Some indicators, including prediction accuracy, time, misclassification level, and F1-total, are considered to calculate the predictor, checking that the REHWX-DNC method proposed is efficient and trustworthy for predicting heart disease.