Stacked Ensemble with Voting-Classifier for Stroke Prediction

Abstract

Stroke is one of the major causes of death nowadays. It occurs rapidly and has severe sequelae, seriously affecting the quality of life of patients. Therefore, early identification of high-risk individuals and intervention can effectively prevent strokes. Machine learning technology has shown great potential in the field of disease prediction. Many medical fields have begun to widely use machine learning to assist in diagnosis. However, there are numerous machine learning algorithms, making it difficult to choose, and different algorithms perform differently on different problems. A single model is unable to balance the advantages and disadvantages of performance to make more accurate predictions. Therefore, this study aims to develop a stacked ensemble model with Voting-Classifier as the meta-model to combine the advantages of models that perform well on this problem, and use real data as the training set to identify high-risk individuals. Moreover, the model performance should be more accurate and more robust than that of a single model.