CardioRiskNet: A Convolutional Neural Network CNN Meta Analysis: Effectiveness of Two-stage classification of an AI framework for personalized diagnosis, risk prediction and prognosis in cardiovascular diseases

Abstract

The leading cause of death worldwide is heart attack, also called cardiovascular disease. Therefore, an accurate and effective early prediction system for heart attack is required. In this paper, a hybrid deep learning model with a combination of Convolutional Neural Network (CNN) and Support Vector Machine (SVM) is proposed to predict heart attack. CNN is used to automatically learn the deep features from the dataset, and SVM is used to perform effective classification with a better hyperplane. Artificial intelligence-based heart attack prevention systems provide a novel approach that has the potential to revolutionize early diagnosis, tailored prevention, and treatment. This work is done by using the UCI Heart Disease Dataset with common preprocessing and hyperparameter tuning techniques. It is observed that the proposed CNN-SVM model outperforms all Machine Learning models and Deep Learning models by showing 97% of accuracy, 94.2% of precision, 92.6% of recall, 93.4% of f1-score, and 0.96 of AUC. This work concluded that the hybrid models are very beneficial for improved accuracy and prediction. There is a possibility that recuperation and therapy won't be necessary under certain circumstances. Further research, development, clinical trials, and interdisciplinary collaboration are required to fully realize the benefits that AI-based heart attack preventive systems offer. These actions would be able to solve the issues that result from this problem.