A Framework For Pregnancy-Induced Hypertension Prediction Using Machine Learning Approach

Abstract

Pregnancy induced hypertension is a common and very severe medical disorder specific to pregnancy and it is a major health burden and one of the leading causes of maternal and perinatal morbidity and mortality. The goal of this study is to design a framework for pregnancy induced hypertension prediction through integrating healthcare server and machine learning algorithm and techniques. As a solution for this problem, this research proposed pregnancy induced hypertension prediction using machine learning and feature selection/extraction method to build a prediction model and used flask server to integrate prediction model with healthcare server. sample data was collected from three hospitals such as Adama hospital,asella hospital and noah speciality clinic and this hospitals was selected based on the largest services given for pregnant women. the sample data was processed and labeled with the help of health experts and classified into three class. The research employed an experimental approach to select the combination of feature extraction and selection method with a machine learning model using RF,SVM and NB. The models evaluated using 10-fold cross-validation, and classification performance is used to compare the models. The study resulted classification result using random forest with combination of feature extraction using PCA resulted with accuracy of 0.97 than SVM and NB. RF was selected to integrate with flask server and evaluated the framework by health experts.