Predictive Power of AI: Tackling Child Undernutrition in Morocco

Abstract

This study examines the efficacy of various machine learning techniques to predict undernutrition in Moroccan children under five, focusing on stunting, wasting, and underweight. Utilizing data from the 2018 Moroccan Population and Family Health Survey, we assessed the performance of machine learning classifiers, including logistic regression, k-nearest neighbors, random forests, and gradient boosting methods. The research reveals that logistic regression and gradient boosting are notably effective in predicting stunting and wasting. Importantly, logistic regression also stands out for its strong predictive accuracy in identifying underweight conditions, making it a valuable tool for public health practitioners. Furthermore, the study delves into the key determinants of undernutrition in Morocco. The research highlights maternal education and household wealth as key undernutrition determinants, pointing to a significant socioeconomic impact. It also notes regional, healthcare, and dietary diversity’s roles in undernutrition outcomes, with rural and lower socio-economic children being most at risk. These insights underscore the importance of tailored interventions to combat undernutrition in Morocco, demonstrating machine learning’s critical role in shaping effective public health strategies.