Exploring Predictors of Stunting Risk Through Machine Learning

¹ Nutritionist Professional Education, Faculty of Medicine and Nutrition, IPB University, 16680 Bogor, Indonesia
² Computer Science Study Program, Faculty of Data Science, Mathematics and Informatics, 16680 Bogor, Indonesia
³ Postgraduate Study Program in Nutrition Science, Faculty of Medicine and Nutrition, IPB University, 16680 Bogor, Indonesia
⁴ Nursing Study Program, Faculty of Medicine, Public Health, and Nursing, Universitas Gadjah Mada, 55281 Yogyakarta, Indonesia
⁵ Department of Child Health, Faculty of Medicine, Bandung 40161, Indonesia

¹ Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

Stunting has long-term health impacts, and short birth length is an early indicator of stunting. The application of Artificial Intelligence (AI) offers opportunities to build more accurate and interactive predictive models. This study aims to develop a classification model to predict the stunting status of children under 2 years old. Model for estimating stunting status by leveraging ML algorithms. The model was created using the XGBoost algorithm, implemented in Python with the scikit-learn module, and executed on Google Colab. Interactive input of attribute values allowed the system to predict lenght-for-age z-score (LAZ) categories. The XGBoost model built in this study was tested on a dataset of 72 samples. The variables showing the highest correlations, based on Spearman’s rank correlation analysis, include the age of children under two, knowledge of complementary feeding (CF) practices, the presence of acute respiratory infection (ARI) symptoms, breastfeeding status, early initiation of breastfeeding (EIBF), maternal body weight, and the frequency of milk consumption. The XGBoost-based ML model achieved 90% accuracy in predicting stunting risk, demonstrating that the AI algorithm is reliable for data-driven stunting prediction.