Enhancing Predictive Accuracy in Cardiovascular Disease Diagnosis: A Hybrid Approach Using RFAP Feature Selection and Random Forest Modeling

Abstract

Cardiovascular disease remains one of the leading causes of death all over the world and is characterized by ailments affecting cardiovascular structures, such as the heart failure, arrhythmia, and coronary artery diseases. It has multiple roots based on genetic influences on the disease, meals, drinks, physical inactivity, or presence of other diseases including hypertension, and diabetes. Heart diseases are still rampant in the world despite the development in the research on diseases and in healthcare. This is mainly because of the increasing average life span of individuals as well as increased cases of urbanization and adoption of poor-quality diets and leading inactive or sedentary lives. Such problems further escalate due to shortcomings in healthcare provision and preventive options by socioeconomic differences, especially affecting the needy populace. Hence, there is a rising focus on using complex technologies such as artifices and machine intelligence, along with big data analytics and predictive models, in handling these issues. These methodologies utilize computational methods and large-scale data to enhance risk assessment, optimize treatment pathways, and alleviate the overall burden of cardiac disease on healthcare systems and individuals. The concentration of this research is on the systematic application of rigorous data preprocessing, class imbalance correction using techniques such as SMOTE, and outlier management using methods such as IQR, in order to improve the effectiveness of predictive modelling. This study intends to improve predictive accuracy and model generalizability, thereby contributing to improvements in healthcare decision-making and patient care outcomes, by utilizing the RFAP hybrid technique for feature selection and Random Forest for modelling.

Topics & Concepts