Heart attack prediction using machine learning and XAI

Abstract

Predictive analytics has received a lot of attention in recent years due to advances in supporting technology, particularly in the areas of big data and machine learning. In recent years, the uses of disease prediction has been seen in the healthcare area. Among so many predictions, this project will show the prediction of a heart attack. Heart disease, often known as cardiovascular disease, refers to a variety of illnesses that affect the heart and has become the leading cause of mortality worldwide in re cent decades. It links a slew of risk factors for heart disease with a pressing need for precise, dependable, and practical methods for making an early diagnosis and man aging the condition. In the healthcare industry, data mining is a typical methodology for analyzing large amounts of data. Because predicting cardiac illness is a difficult undertaking. It is necessary to automate the process in order to avoid the risks connected with it and to inform the patient well in advance. Heart diseases can be determined using data mining techniques such as XGBOOST, Logistic Regression, Stochastic Gradient Descent, Support Vector Classifier, Kneighborsclassifier, and Naive Bayes. With this project, I have shown that among all the above machine learning models, XGBOOST outperforms other techniques in terms of predicting heart attacks. As a result, this paper conducts a comparative study of the perfor mance of several machine learning algorithms. For any type of prediction features of the dataset plays a very important role. Features can give positive or negative impact on the final prediction. The features importance can be visualized by the XAI methods. This paper also takes an approach to interpret the explainability of the model’s prediction. By using the XAI method SHAP and LIME with the help of the concept of black box, this research conducts the KNN algorithms prediction.