Enhancing Cardiovascular Disease Prediction with XAI Technique Using Machine Learning

Abstract-Globally, coronary diseases (CV) are several of the most significant causes of demise, improvements in predictive healthcare technologies are imperative. The goal of this study is to improve the predictability and interpretability of cardiovascular disease prediction models by combining machine learning methods with Explainable Artificial Intelligence (XAI). To create reliable predictive models, we investigate a range of machine learning algorithms, such as ensemble approaches, logistic regression, and XG-Boost. But while though precision is crucial, these predictions’ interpretability is just as crucial for therapeutic use. Our goal is to make model procedures for making decisions concise and intelligible for physicians by utilising XAI techniques like SHAP (Shapley Additive Explanations) and LIME (Local Interpretable Model-agnostic Explanations). Using a real-world CVD dataset, our tests demonstrate that XAI-enhanced models do not not only increase the accuracy of predictions but also identify important variables affecting heart function. By providing a workable framework for using interpretable machine learning models in healthcare, this study advances the discipline and may result in better clinical judgements and more individualised patient care.The accuracy of the Random forest-CARDIO system is assessed against the Framingham heart disease dataset using the Colab Simulator. In the experiment, Random forest demonstrated a significant accuracy score of 91.38%, which is appreciably better than alternative techniques including, XGBoost (90.01%), RNN (85.02%), GRU (85.02%) and RNN+GRU (as a combined model) (86%).