Cardiovascular disease is regarded as one of the main sources of death for the entire planet. Drawing the conclusion of cardiovascular failure is a difficult task, especially in immature and agricultural countries that lack human experts and types of equipment. Since then, various experts have created various intelligent frameworks to mechanize the identification of cardiovascular diseases. Feature importance/selection is a key part of the medical data set. In this article, we propose a diagnostic system that uses chi2, logistic regression (LR), Pearson, recursive feature elimination, and random forest (RF) as feature selection and several classifiers to predict heart disease. Among the 11 features in the heart disease data set, important features were selected. Apply accuracy and other measures such as precision, sensitivity, specificity, F1 score, ROC (receiver operating characteristic), Log_Loss, mathew_corrcoef, and confusion matrix to compare the data set with all features and selected features. Experimental results show its effectiveness and effectiveness in predicting heart disease. In addition, the proposed model shows better performance compared to the previously proposed model. In addition, our proposed method achieves a high prediction accuracy of 82.95%. Our results show that the proposed method can be reliably used to predict clinical heart disease.