Improving Cardiovascular Prediction with Advanced Machine Learning and Deep Learning Approaches

The World Health Organization (WHO) reports that cardiovascular illnesses are the primary killer worldwide. Estimates for 2019 indicate that cardiovascular disease will be responsible for the deaths of 17.9 million individuals worldwide. Heart disease is more common in high-income nations because of factors like inactivity, poor nutrition, and an aging population. However, urbanization and improvements in living standards are also contributing factors in the rise of heart disease in developing and middle-income nations. Predicting and diagnosing cardiac disease with Machine Learning (ML) and Deep Learning (DL) algorithms has shown encouraging results. In order to find people at risk for or already suffering from heart disease, these algorithms make use of patterns and linkages within massive datasets. The heart disease dataset from the UCI repository is used and many distinct ML and DL algorithms applied to maximize prediction precision. The following are the findings of comprehensive analysis of the algorithms, which determine their trustworthiness and compatibility by contrasting their individual and combined performance. Prediction accuracies of 98.53% were attained by XGBoost, 89.61% by AdaBoost, 98.53% by Decision Trees (DT), 88% by Support Vector Machines (SVM), 98.54% by Multi-Layer Perceptron (MLP), 90% by Artificial Neural Networks (ANN), 98.53% by combining AdaBoost and XGBoost, 83% by K-Nearest Neighbors (KNN), and 93.65% by using all four (KNN, SVM, ANN, DT) of these methods together.