Research on predictive algorithms for cardiovascular disease

Abstract

Cardiovascular disease (CVD) is a global disease with acute and chronic complications. It is primarily responsible for the vast majority of deaths worldwide, which account for 17.9 million deaths annually. In terms of CVDs, illnesses like rheumatic heart disease and coronary heart disease are included, of which coronary heart disease (CHD) accounts for more than 50% of all these cases. In this research, principal component analysis (PCA) and backward stepwise elimination are used to identify the relevant predictors and avoid overfitting models for random forest analysis and logistic regression analysis. Moreover, for assessing the effectiveness of the models, the confusion matrix and the receiver operating characteristics (ROC) curve with AUC (area under the ROC curve) value are produced for model comparison. The outcomes demonstrate that the random forest model performs better at categorizing high-dimensional data. Thus, the techniques discussed in this paper give medical researchers better ways to handle coronary heart disease data statistically and provide a new statistical procedure for coronary heart disease prediction and prevention.