Efficient Prediction of Cardiovascular Disease by Fusing Boosting Classifiers Combined with Explainable Approach

According to the World Health Organization (WHO), the leading cause of mortality in the modern world is cardiovascular disease (CVD). This disease is responsible for the death of 17.9 million people worldwide, accounting for 31% of all deaths. This research introduces a Machine Learning (ML) based framework for physicians, in which no patient information is disclosed and physicians receive a prior prediction based on the existing data with explanations. In our work, we propose an advanced framework that initially finds the individual prediction for five tree-based boosting models utilising 10-fold cross-validation. Then, after concatenating the predicted values for each boosting model along with the data set, Random Forest has been exploited to provide the final prediction. Besides, to balance our data, we have used the up-sampling method and Robust Scaler to eliminate outliers. We have additionally added Explainable-AI generated figures to enhance the trustability of the prediction. The result shows the prediction is pretty accurate in terms of evaluation metrics (82% accuracy with 81% precision, 82% recall, 82% ROC-AUC Score, and 80% F1-Score) compared to other recent states of the art research in this domain by which we can save a lot of human lives.