ACCURACY ASSESSMENT OF MACHINE LEARNINGALGORITHMFOR PREDICTION OF HEART STROKE

Abstract

This paper conducts a relative analysis of machine learning algorithms applied inthe prediction of heart strokes. It is necessary to automate the heart stroke prediction

procedure because it is a hard task to reduce risks and warn the patient well in

advance. The cardiac stroke dataset is used in this work. The suggested work uses various approaches including Decision Tree , Random Decision Forest , Extreme Gradient Boosting,  and Hybrid model to forecast the likelihood of Heart Stroke and categories patient risklevel. These algorithms are crucial in automating the prediction process, leading to enhanced patient outcomes and greater efficiency in healthcare delivery. The core of our investigation revolves around assessing the execution of these algorithms in respect of predictive precision, sensitivity, specificity, and other relevant grade. Our methodology includes meticulous data preprocessing, model training, and rigorous evaluation using standard healthcare performance metrics. The results of our testing reveal that, among the various machine learning algorithms considered, the RF approach exhibits the greatest level of accuracy. This suggests that Random Forest has the potential to be a valuable tool in early heart stroke prediction and risk stratification. The testing outcomes indicate that among the ML algorithms utilized, the RF approach demonstrates the highest accuracy..