Cardiovascular Health Risk Prediction Using Bayesian Predictive Analytics Approach

A correct CVD diagnosis and outcome, on the other hand, result in expedited patient care and highly accurate treatment, as well as good results. Medicine has turned to machine learning since it is capable of discerning intricate patterns from vast data set insights. This study seeks to reduce the risk of miscategorizing CVD events and subsequently decrease CVD death rates by using machine learning methods. As a result, a more sophisticated approach outlined: RF, Decision Tree, MP, and XGB as the best current computer program in the proposed system. To refine the sorting capability of the model, a revolutionary method produce, including a K-modes grouping and a Huang starting point technique to increase sorting the stuff more effectively. Using an actual data from Kaggle dataset which had 4240 examples, our models were trained and tested. After rigorous evaluation, the models demonstrated noteworthy accuracy rates: Decision Tree achieved 72.81%, XGBoost exhibited 83.33%, Random Forest attained 83.61%, Naive Bayes achieved 81.01%, and Multilayer Perceptron outperformed others with an accuracy of 77.32%. Moreover, effectiveness of the model was calculated by using the Area Under the Curve (AUC), which resulted in outstanding predictive capabilities. Specifically, XGBoost and Random Forest both achieved an AUC of 0.58, with support vector machine (SVM) model outperforming all, boasting an accuracy level of 83.33%, an AUC score of 0.58, highlighting their robustness in predicting CVD occurrences. An observation from this study tells the superiority of the brain-like network model (MPM), particularly at the time matched with cross-checking. It resulted in achieving an exceptional success rate of 87.28%. The research developed not only focuses on the area of predicting heart diseases but also lays emphasis on the potential of machine learning methods to transform medical diagnostics.