Classification and Prediction of Cardiovascular Disease Using Machine Learning Techniques

An investigation indicated that one out of every four deaths is caused by cardiovascular disease. Early detection of some illnesses can help save lives, and increasing awareness of heart health can do just that. It has been proven that cardiology specialists can predict the likelihood of a heart attack with about 67% accuracy. Our ultimate objective is to create a machine-learning-based model that is more precise than human doctors. Some of the machine learning methods that can improve our confidence in our predictions are logistic regression, random forest, support vector machine, and ensemble learning. Diagnosis accuracy of physicians could be considerably enhanced by a model of classification for cardiac sounds that can be used by any physician, radiologist, or patient. The model classifies the heartbeats as normal, murmur, and extrasystole based on the S1 (or lub) and S2 (or dub) heartbeats. To separate the most crucial heartbeats, a segmentation method is constructed. Without the need for expensive and time-consuming specialized tests, that may assist doctors in rendering fast suggestions to their patients. By connecting a sound recorder to the stethoscope, the process can be streamlined and made more accurate. In this study, we present the accuracy rates of various algorithms for heart disease prediction and heart sound classification, demonstrating the potential of Logistic Regression, Random Forest, Ensemble Learning, Support Vector Machine, and Convolutional Neural Network in enhancing diagnostic outcomes for cardiovascular health. In this manuscript Support Vector Machine demonstrated the highest accuracy of 90.11%, showcasing its proficiency in identifying cardiovascular data patterns. Logistic Regression also performed well with an accuracy of 86.90%. Additionally, the success of CNN in heart sound categorization, achieving 85% accuracy, highlights the effectiveness of deep learning in appropriately classifying cardiac sounds. Furthermore, the adaptability of Support Vector Machine, with a 70% accuracy, in both heart illness prediction and heart sound classification is commendable.