MACHINE LEARNING MODELS FOR CLASSIFICATION AND PREDICTION OF PREECLAMPSIA IN KADUNA, NIGERIA

Abstract

Preeclampsia is a significant complication in pregnancy characterized by high blood pressure and damage to organs, posing serious risks to both the mother and fetus. Early prediction and management are crucial for improving outcomes. In Nigeria, where healthcare resources are often limited, and prenatal care access can be uneven, advanced predictive models can enhance early detection and intervention. This paper developed machine learning-based classifiers and predictors of preeclampsia. The data used was collected from general hospital Hunkuyi, Kaduna state, Nigeria. The models were based on Adaboost, Support Vector Machine (SVM), and Naïve Bayes (NB) algorithms.  In both classification and prediction of preeclampsia among the population studied, SVM has 0 error in MAE, RMSE, RAE and ERSE, with accuracy level of 100%. Adaboost and NB had accuracy levels of 98% and 85%, which are very good. This paper recommends the use of these models for prediction of onset of preeclampsia among pregnant women in Hunkuyi and Kaduna state. Since the data used to develop the models represent impartially the various set of people within the state, it can be used for all women. we believe the models can assist the health personnel to predict onset of preeclampsia and help proper planning and intervention. It will also reduce maternal and child mortality that could result for preëclampsia.