Ensemble learning as a prerogative method of predicting mortality of patients with cardiovascular diseases

Abstract

Among diseases that account for higher death rates, Cardio Vascular Diseases (CVD) stand forefront. Many works have been carried out since long to predict effectiveness in mortality prediction using models like data mining, logistic regression, neural networks etc. considering only traditional cardiovascular risk factors. As time and technologies evolved with incorporation of newer features these models ended up with predicted mortality rate of an accuracy 60-70%. There are many more attributes to be explored that are significant in predicting mortality rate in CVD patients, opening the scope to develop prediction models with traditional and non-traditional risk factors, much wider. This paper is focused on predicting mortality rates using three models. Each model's performance metrics are calculated to check the accuracy of the model. This helps one to build models that could best predict the outcome. Use of Ensemble learning method enhanced the prediction accuracy to 91%. This helps to validate the decision more accurately about mortality predictions and thereby assessing the risk.

Ähnliche Arbeiten