Performance Analysis of Machine Learning Algorithms for Hypertension Decision Support System

Abstract

Machine learning algorithms are helpful to build a model-based decision support system using data to predict risk of hypertension disease which is deadly in Bangladesh as in other parts of the world. It is necessary to figure out which machine learning algorithm is suitable for implementing a decision support system practically. Therefore, in this work, 21 types of supervised machine learning algorithms have been employed training the prediction system for hypertension risk. Various types of Decision Trees, Logistic Regression, Support Vector Machines, Nearest Neighbors Classifiers and Ensemble Classifiers are used for training the model. 5 fold cross validation has been used in this case. 16 inputs are chosen based on expert knowledge and 2 outputs are selected as response. In this paper, performance is evaluated in terms of confusion matrix and ROC curve. 129 patients' data have been collected from local hospital to conduct this work.