Enhanced AI-Powered Coronary Artery Disease Prediction Using Spider Swarm Intelligence With Deep Hyper Scalar Capsule Gated Recurrent Neural Network for Early Treatment Recommendation

Resumo

Nowadays, increasing numbers of heart attacks are commonly suffered by young age people due to unhealthy lifestyles. Especially Cardiovascular Disease (CVD) or coronary artery disease (CAD) are the most dangerous, deadliest diseases. Without knowing their symptoms, heart block attacks and lead to death. Unhealthy lifestyles are an essential aspect of this disease, which is analyzed by healthcare sectors through behavior and symptoms to identify with support of CT or angiogram support. So many traditional methods are implemented in Machine Learning (ML) and Deep Learning (DL) methods to identify heart disease by analyzing the disease's relational features and behaviors. Most of the techniques concentrate on disease margins to select non-relational features. So, the analysis degrades the accuracy of the performance and increases the false positive rates. Consequently, the precision, recall, and f1 attain poor performance. To resolve this problem, we proposed an advanced Artificial Intelligence-Powered Deep Hyper Spectral Learning Technique for heart disease prediction using Optimal Spider Swarm Intelligence Technique (OSSIT) Feature Selection with Deep Hyper Scalar Capsule Gated Recurrent Neural Network (DHSC-GRNN) for early prediction and recommendation. Initially, the Min Max normalization is applied to normalize the CAD heart disease dataset, and the Cardiac Feature Behavioral Impact Rate (CFBIR) is analyzed through decision tree ideal margins. By reducing the non-relational features and introducing the OSSIT, it is applied to select the mutual dependencies of feature patterns. Finally, the DHSC-GRNN is used to predict heart disease by class by risk. The feature patterns are scaled into capsule units from a related class cluster and fed into recurrent gated units to reduce non-scaling margins and disease severity resemblance to predict the risk. The proposed system achieves high performance compared to the other system in terms of higher precision rate and recall rate f1 measure by increasing the accurate favorable rates compared to the preliminary traditional methods.