HEART DISEASE PREDICTION USING CARDIO CAPSULE NEURAL NETORK

Abstract

            Heart disease is a major public health problem across the world, and precise prediction of cardiovascular events is critical for preventative healthcare. This research suggests a complete method for heart disease prediction that includes data preparation, feature selection, grouping, and training using specialized neural networks. The first step is to use the Weighted Transform K-Means Clustering technique to improve data quality by eliminating noise and expanding the dataset. This preprocessing approach helps to create a more resilient and trustworthy dataset for further analysis. To select the most significant characteristics for heart disease prediction, an Ensemble Feature Selection technique is used in conjunction with the Weighted Binary Bat Algorithm. This approach aids in determining a selection of characteristics that substantially contribute to the classification problem. Clustering similar instances together is critical for identifying patterns in the data. As a result, Density-Based Ordering of Clustering Objects is used to group instances with comparable characteristics, revealing possible risk factors and linkages within the dataset. The last level of training and classification uses Cardio Capsule Neural Networks (CCNN). CCNNs are specialized neural networks created exclusively for medical purposes, with an emphasis on cardiovascular prediction tasks. The model is trained on the preprocessed and chosen features, which allow it to discover complicated patterns and connections within the data.