Published in PLOS One on 2026-01-02, this study is presented as an attributed evidence briefing. The excerpts below are the original researchers’ words, reproduced under their Creative Commons Attribution licence. Health Signal has selected the abstract and, where needed, opening background paragraphs; the complete methods, results, limitations and references remain in the linked paper.
Original authors: Halid Worku Jemil, Sonia Worku Semayneh, Altaseb Beyene Kassaw, Kassahun Dessie Gashu, Olutosin Otekunrin
Source: PLOS One • DOI 10.1371/journal.pone.0340221
Read the complete primary study ↗ · Creative Commons Attribution licence
From the researchers
Abstract
Severe stunting is one of the primary public health challenges in LMIC including Eastern African Countries, which affects millions of children. In addition, it was a major contributor for mortality and related complication of children aged under five. However, there is limited study conducted severe form of stunting by employing Machine learning (ML) in Eastern African Countries. Therefore, our study was demonstrated to predict and identify its major determinants using ML algorithms, furthermore, to improve model explainablity. Our study used Shapley Additive explanations (SHAP) and ARM to identify the determinants of severe stunting among under-five.
cross-sectional study was conducted using DHS data from 2012–2022 in East Africa. 136,074 children were the source populations, and 76,019 children were the study population. Data were analyzed using Python version 3.7 and R version 4.3.3 for data preprocessing, modeling, and statistical analysis. Model performance was evaluated using accuracy and AUC. Furthermore, the SHAP analysis and ARM was used to further explain and interpret the determinants of severe stunting among children under five.
The Random Forest performed the best in this analysis, with an accuracy of 87% and an AUC score of 0.83. The analysis indicated that women’s who do not practicing exclusive breastfeeding (SHAP value = +0.41), being from Burundi (SHAP value = +0.04), children being underweight (SHAP value = +0.25), lived in poor household (SHAP value = +0.40), child gender being male(SHAP value = +0.23), mothers height being short (SHAP value = +0.03), mothers being underweight (SHAP value = +0.18), child size at birth being small (SHAP value = +0.21), women’s being delivered in home(SHAP value = +0.07), mothers education being primary (SHAP value = +0.20), unimproved toilet (SHAP value = +0.06), distance to health facility being a big problem (SHAP value = +0.02), were associated with increase the risk of severe stunting among under five.
The Random Forest was the best-performing model for predicting severe stunting in Eastern African countries. To decrease the effects of severe stunting, integrated interventions should provide support for mothers with lower socioeconomic conditions, strengthen maternal education, empower women to practice exclusive breastfeeding, encourage facility deliveries, increase access for households to sanitary facilities, provide education on personal and environmental hygiene, provide mothers with information on the importance of complementary feeding for children as well as for the mothers, and provide near health facilities for mothers and essential care services.
Introduction
Severe stunting is a critical form of malnutrition, causing extreme growth impairment and development in children. It is mainly caused by a chronic lack of adequate dietary intake, repeated infection, and inadequate psychological stimulation, resulting in several health and developmental problems [1–3]. It results in destructive health consequences, including impaired defense against infection, growth retardation, delayed recovery from infection, poor intelligence quotient, high anxiety and depression, and increased susceptibility to disease. It also causes poor economic productivity and poor educational attainment [4–7].
Reading this evidence
Use the full paper to check the study design, participant selection, comparator and outcome definitions before drawing a practical conclusion. An association does not by itself show cause and effect. A reported performance measure needs its denominator and uncertainty. A finding from a particular facility, sample or time period should retain that scope when it is discussed elsewhere.
This briefing makes the source accessible within the wider technology & industry desk. It does not assess an individual patient or certify a product. When considering relevance to another health system, examine the original authors’ account of context and limitations, then seek local implementation evidence. The related reporting below provides other studies to read alongside this one; it is a topic connection, rather than a claim that their results are directly comparable.
© 2026 Halid Worku Jemil, Sonia Worku Semayneh, Altaseb Beyene Kassaw, Kassahun Dessie Gashu, Olutosin Otekunrin. Excerpts reproduced under CC BY. Selection, formatting and contextual notes by Health Signal. The researchers are the authors of the source study, not members of Health Signal’s staff.
Explore connected evidence
- Evidence brief: Machine learning-based prediction of fever among under-five children in Ethiopia: A national-level study
- Evidence brief: Beyond health: A machine learning analysis of structural barriers to school attainment in Somalia
- Evidence brief: Exploring the association between home delivery and under-five mortality in Ethiopia: A bivariate multilevel mixed-effects analysis
