Prediction of cardiovascular disease by using a genetic algorithm over a stacked ensemble model

A two-layer customized ensemble model is suggested in this work to improve the prediction accuracy of cardiovascular disease. Two ensemble classifiers are stacked as part of the model architecture. While Support Vector Machine (SVM) and Decision Tree (DT) are used as base learners in the first ensemble, and Random Forest (RF) and Logistic Regression (LR) are used in the second ensemble. The dataset used in training the model has been collected from the UCI data repository. The most pertinent attributes were chosen using LASSO (Least Absolute Shrinkage and Selection Operator), which lowers dimensionality. To further enhance the model’s performance, the Genetic Algorithm (GA) has been used. The suggested model’s excellent efficacy in detecting cardiovascular disease has been demonstrated by its 88.52% accuracy rate. 10 Fold Cross Validation has been added for further validation of the model and it has also been identified statistically significant over Decision Tree and Support Vector Machine classifier as per the tested results. For better understanding of prediction the proposed model has been explained using SHAP. With a strong and precise prediction framework for medical practitioners, this hybrid ensemble technique shows great promise in medical diagnosis systems.