Improving Prediction of Mortality in ICU via Fusion of SelectKBest with SMOTE Method and Extra Tree Classifier

Heart disease is one of the most prevalent and serious health widespread issues affecting elderly and middle-aged individuals globally. Cardiovascular diseases (CVDs) impose considerable morbidity and mortality rates and entail considerable financial strain on global healthcare infrastructures. According to the report of the World Health (WHO) Organization, the mortality rate of heart disease will increase to 23 million cases by 2030. In healthcare, predicting diseases and analyzing electronic health records to derive useful patterns aid in early and accurate CAD diagnosis. Hence, in this paper, we worked on the Z-Alizadeh Sani dataset to demonstrate the strong ability of the machine learning technique in predicting CAD. We also applied the genetic algorithm to reduce dimension by finding the important features of the neural network. The results showcased that our proposed method could diagnose CAD by achieving the highest accuracy, sensitivity, and AUC of 94.71%, 96.29%, and 93.5%, respectively.