Predicting Diarrhoea Among Children Under Five Years Using Machine Learning Techniques

Abstract

Globally, diarrhoea remains a significant cause of death among children under five years. Several preventive interventions such as hygiene practice, safe drinking water, rotavirus vaccination and health promotion were implemented to reduce the catastrophic impact of diarrhoea. However, effective tackling of the diarrhoeal disease requires robust preventive interventions and computational techniques to predict diarrhoea among children under five years using risk factors. Therefore, this study applied a decision tree classifier, logistic regression and support vector machines to predict diarrhoea among children under five years using the recent Zimbabwe Demographic Health Survey dataset. The study revealed that logistic regression out-performed other diarrhoea predictive models with the prediction accuracy of 85%, precision of 86%, recall of 100% and the F1-score of 94%. Support vector machines also performed better in predicting diarrhoea with predicting accuracy of 84%, precision of 85%, recall of 100% and F1-score of 92%. The study also revealed that understanding risk factors such as climatic or meteorological, socioeconomic and demographic factors plays a tremendous role in tackling diarrhoea among under-fives. The application of machine learning techniques can assist policymakers in designing effective and adaptive diarrhoea preventive interventions, control programmes and strategies for tackling diarrhoea.