Early Prediction of Cerebrovascular Disease using Boosting Machine Learning Algorithms to Assist Clinicians

Abstract

Clinicians are required to make an early prediction of diseases to save a  life, especially cerebrovascular diseases. The objective of this research is to use mathematical models such as boosting machine learning algorithms as a tool to be applied by clinicians for cerebrovascular disease. This paper particularly, considered XGBoost, AdaBoost, LightGBM, and CatBoost Classifiers to predict cerebrovascular disease using age, gender, BMI, hypertension, heart disease, residence type, ever married, smoking status, and average glucose level of the patients. Synthetic Minority Over-Sampling Technique Edited Nearest Neighbors Under-sampling (SMOTE-ENN) and Feature Engineering were applied to the dataset to enhance the performance of the algorithms. The result obtained showed that XGBoost Classifier is the best model with an accuracy of 98% and an AUC of 0.983.