Cardiovascular Disease Prediction with Machine Learning Algorithms and Interpretation using Explainable AI methods: LIME & SHAP

Every year a third of the population dies due to cardiovascular disease and these numbers grow exponentially which is one of the leading cause of deaths. Cardiovascular disease is a group of disorders that affect the heart and blood vessels. These disorders often result from high cholesterol, high blood pressure, smoking, obesity, and many other medical attributes. Therefore, early intervention and diagnosis of cardiovascular diseases improves the patient’s quality of life. In this era of AI and ML, the ML techniques integrated with explainable AI are a promising tool for implementing and getting improved accuracies in the detection of cardiovascular disease. With the usage of Binary classification dataset on heart disease this paper explored a comparative study with various machine learning algorithms which includes single classifiers (LR, KNN, DT, SVM), bagging(RF), and boosting(XG, ADA, CAT, GRADIENT), ensemble machine learning model using RF, GRAD, ADA, XG, CAT helped in achieving exceptional accuracy of 89.13%, ensemble model with hyperparameter tuning techniques (GridsearchCV and RandomsearchCV) and implementation XAI methods(lime and sharp). In a comparative analysis of boosting algorithms specifically CAT and ADA boost outperformed other algorithms in identifying cardiovascular disease. The hyperparameter tuning techniques help improve the accuracy of Random forest classifier method to an impressive 88.26%. Additionally, the use of XAI methods such as LIME and SHARP provides global and local insights into model behavior and feature importance and represents it graphically which helps in better understanding.The use of these machine learning models in remote areas where diagnosis and specialized treatment of heart-related disease is limited this model plays a pivotal role by helping in early intervention and diagnosis of cardiovascular disease.