Revolutionizing HD Diagnosis: Hybrid Deep Learning and Machine Learning Using ISV-CNN

Abstract

Millions of people in worldwide suffers from Heart Disease (HD), in which Coronary Artery Disease (CAD) is the most prevalent kind. Timely and accurate diagnosis is essential for successful therapy and better patient outcomes. This study introduces a novel Integrated Stochastic Vanilla Convo-Neuro Net (ISV-CNN) method to enhance the accuracy of HD detection by analysing complex data patterns. The research also utilized a publicly available dataset that was pre-processed using Z-score normalization and Independent Component Analysis (ICA) for feature extraction. The ISV-CNN model considers the temporal dependencies for identifying HD, while the Particle Swarm Optimization (PSO) is employed to optimize the performance metrics. The proposed approach was implemented using Python tools, and its best attain performance was evaluated based on the accuracy (98.5%), precision (98.98%), recall (98.7%), and F1 score (98.8%), Sensitivity (98.7%) and Specificity (98.9%). The ISV-CNN model demonstrated significant improvements over the existing methods, particularly in detecting HD, where temporal trends are important. The incorporation of PSO for optimization and ICA for feature extraction further contributed to the effectiveness of the model. The ISV-CNN approach offers a comprehensive solution for HD identification by efficiently analysing complex data and temporal relationships, outperforming the current diagnostic methods. These findings underscore the potential of integrating domain-specific knowledge with machine learning techniques for the early diagnosis and effective treatment of HD, ultimately improving patient care and outcomes.