Effective Heart disease prediction framework using Random Forest and Logistic regression

Abstract

Heart or cardiovascular disease claims approximately 17.9 million lives each year and has become a significant cause of death. Prior detection of CVDs (cardiovascular diseases) could determine dietary and lifestyle alterations needed for a high-risk patient's life, which will lessen the challenges they might face in the future. Several people experience cardiovascular conditions (CVDs) due to their current lifestyle which have even resulted in fatalities across the world. This research has attempted to narrow down potential factors for cardiovascular disease as well as reliably estimate all in all danger. Applying uniform methods in data mining such as logistic regression and random forest will help to accurately determine the factors which play a role in it. To acquire the best accuracy and to understand the parameters of the application, the LR model is first given training with five splitting conditions and then assessed with a training dataset. The algorithm generates categories of one and zero for heart disease occurrence and avoidance, respectively. Significant information is enabled by it, such as patterns and linkages. The key topic of this research work is centered on the patients, given specific health parameters, who are more likely to suffer heart problems. Using the patient's medical records, we put together a method that can identify if a cardiovascular disease assessment is required or not for the individual. To detect and label an individual with heart problem, we employed two different machine learning methods i.e., logistic regression and random forest. The precision value obtained by the Logistic Regression method is 85.25%, and the precision value obtained by the Random Forest method is 90.16%.