Leveraging Machine Learning Techniques for Improving Heart Disease Prediction Systems Using Feature Selection

Abstract

Although there have been advancements in the Indian healthcare system over the past few decades, there is still a long way to go before we can claim to have reached world standards. Despite being the second most populated nation, India ranks 143rd out of 195 nations in terms of healthcare infrastructure. Even now, seven decades after India's declaration of independence, the country's healthcare system remains unable to guarantee universal access to care. Access to affordable, high-quality healthcare is still a pipe dream, especially for those who live in rural areas. Not everyone can afford medical services. Private organizations charge a high price for their therapies. No considerable financial assistance is allowed. This study suggests a novel hybrid feature selection method for identifying the most important qualities. Standard feature selection approaches such as Maximum relevance and minimum redundancy (mRMR), Relief, a genetic algorithm, and Least absolute shrinkage and selection operator (LASSO) were compared. A variety of classifiers were used to create a cardiovascular disease prediction system, including logistic regression, Naive Bayes, Random Forest, and support vector machine. This study made use of data from the Cleveland heart disease dataset. According to the results of this research, a Random forest based prediction model trained using characteristics discovered via a new hybrid feature selection may provide the best accuracy and sensitivity. According to the results of the research, applying feature selection algorithms enhances the performance of the prediction system in terms of accuracy, sensitivity, specificity, and throughput.