Predict Cardiovascular Disease Using Machine Learning Techniques

Abstract

The declaration emphasizes the increasing prevalence of physical illnesses, especially cardiovascular disease, and the demand for efficient methods to forecast the illness and increase early detection. The utilization of effective algorithm models is recommended by this study in order to predict the risks related to cardiovascular disease. The three datasets used in the study were obtained from the Kaggle website and the UCI machine learning library. Several algorithm models were used in this work, including Gradient Boosting (GB), K-Neighbors Classifier (KNN), XGB Classifier (XGB), Random Forest (RF), Logistic Regression (LR), and Decision Tree (DT). To assess the performances, ensemble models including Bagging, Boosting, Random Subspace, Stacking, and Voting were also used. For the Hungarian dataset, the conventional model KNN earned the highest accuracy with 80.26%, the LR model achieved the greatest accuracy with 88.52%, and the DT model reached the best accuracy with 72.99%. For these three datasets, we also used five distinct kinds of ensemble methods. The findings showed that for the Hungarian dataset, KNN had the greatest accuracy at 81.26%, for the Cleveland dataset, Bagged GB had the best accuracy at 91.8%, and for the Cardio dataset, GB had the best accuracy at 73.11%. By applying hyperparameter tuning, the best parameters were assigned to each classifier, resulting in more precise cardiovascular disease predictions. Comparing the experimental examination to earlier research, performance was better, with the ensemble model obtaining the greatest accuracy of 91.8%. The study emphasizes how crucial it is to use ensemble methods and sophisticated algorithm models in order to forecast cardiovascular illness properly and enhance real-world results.