Digital Healthcare System Using Stacked Ensemble Machine Learning Model to Predict Heart Diseases

Cardiovascular diseases (CVDs) have emerged as the world’s most deadly disease in recent years and are the primary reason of death worldwide. An intelligent and improved system is required to detect heart disease in its early stages accurately, reliably, and in a faster way so that human life can be saved before reaching its critical stage. Many researchers have proposed solutions using a machine learning approach, but need improvement in terms of accuracy and other QoS parameters. Hence, we have proposed two layers of Stacking Ensemble Model (level 0 and level 1) for heart disease prediction that will divide the dataset based on age and gender. From the input training dataset, level 0 models gets trained and the output of this model works as an input for the level 1 model to perform prediction. The experimental results demonstrate that the proposed Stacked Ensemble Model improved accuracy, sensitivity, F1-Score, specificity, recall, and precision parameters. The proposed prediction model attains an accuracy of 95.121% for the male group, and 100% for the female group, which is the best among all the used models.