Cardiac Vascular Disease Prediction Using Stacked Ensemble Algorithm for Type-2 Diabetes Mellitus

Cardiac vascular disease (CVD) is a most serious disease associated with highest humanity and complaint rates in the global population. The risk of developing CVD consciously increases in patients having long-term diabetes difficulties. Consequently, it is necessary to predict the association between CVD and type-2 Diabetes Mellitus (T2DM) for effective decision making to initiate appropriate treatment of T2DM patients. First, the proposed research work to examine the purpose of machine learning (ML) techniques to predict the risk of death rate in CVD patients with T2DM. Then the proposed method is mainly focusing to predict the survival rate of regularity of CVD patient's incidence with T2DM. However, the dataset is unbalanced and handling such unbalanced dataset is one of the major challenges addressed using cost sensitive learning algorithms for reducing the misclassification rate. This proposed method stacked ensemble with cost-sensitive learning that combines the prediction results of the three base classifiers to predict the survival rate of patients with cardiovascular disease and T2DM. The proposed stacked ensemble algorithm uses a 10 fold cross validation method with network search optimization to tune the hyper-parameter of the meta classifier model. The proposed algorithm is used to measure the performance of dataset being evaluated by splitting the data into 80% training and 20% testing. The analysis of outcomes displays that the survival rate of heart failure patients is better predicted using the cost-sensitive stacked ensemble learning technique to achieves precision of 93% and higher accuracy. The unbalanced class and cost sensitive learning may help to address the class imbalance issue. The incorrect weight assignment may lead to overfitting to the minority class while underperforming on the majority class.