Early prediction of chronic heart disease with recursive feature elimination and supervised learning techniques
Year: 2023
Type: article
Abstract: Chronic heart disease (CHD) is a common complication among patients suffering in the cardiological intensive care unit, often resulting in poor prognosis and high mortality. Early prediction of CHD can reduce mortality by preventing the severity of the disease. This study evaluated the efficacy of on recursive feature elimination for predicting CHD using supervised learning techniques for predicting CHD. The study employed 1190 Cleveland Hungarian CHD dataset. Different supervised learning techniques (support vector machine, decision tree, k-nearest neighbor, Naive Bayes, stochastic gradient descent, adaptive boosting, and multilayer perceptron) were used to study the efficacy of the recursive feature elimination. Chest pain type, sex, blood sugar level, angina, depression, and slope were associated with CHD occurrence. The accuracy of the K-nearest neighbor and decision tree model was 89.91% for the feature-selected dataset indicating good predictive ability. Ultimately, the support vector machine and logistic regression with the selected features exhibited good discriminatory ability for early prediction of CHD. Thus, the recursive feature elimination is a good approach to develop a a model with higher accuracy to predict CHD.
Institution: Injibara University
Language: English
Subfield: Health Information Management
Field: Health Professions
Domain: Health Sciences
Keywords: chronic heart disease, early prediction, recursive feature elimination, feature selection, supervised learning, support vector machine, decision tree, k-nearest neighbors, naive Bayes, multilayer perceptron, adaptive boosting, logistic regression, Cleveland heart disease dataset, chest pain, blood glucose, angina pectoris
Open Access status: diamond
OpenAlex ID: W4390176983
