Yehuala Tirualem Zeleke, Baykemagn Nebebe Demis, Terefe Bewuketu
Department Health informatics, Institute of Public Health, College of Medicine and Health Sciences, University of Gondar, Gondar, Ethiopia.
Department of Community Health Nursing, School of Nursing, College of Medicine and Health Sciences, University of Gondar, Gondar, Ethiopia.
Front Public Health. 2025 Jul 23;13:1513922. doi: 10.3389/fpubh.2025.1513922. eCollection 2025.
Diarrhea is the leading cause of childhood malnutrition. Although replacement, continued feeding, and increasing appropriate fluid at home during diarrhea episodes are the cornerstones of treatment packages, food and fluid restrictions are common during diarrheal illnesses in Africa. To fill the methodological and current evidence gaps, this study aimed to build models and predict determinants to increase feeding practices of children in East Africa during diarrheal outbreaks.
We used the most recent demographic and health survey (DHS) statistics from 12 East African nations collected between 2012 and 2023. The analyses included a total weighted sample of 20,059 children aged 5 years. Python software was utilized for data processing and machine learning model building. We employed four ML algorithms, such as Random Forest (RF), Decision Tree (DT), XGB (Extreme Gradient Boosting), and Logistic Regression (LR). In this work, we evaluated the predictive models' performance using performance assessment criteria such as accuracy, precision, recall, and the AUC curve.
In this study, 20,059 children aged 5 years were used in the final analysis. Among the proposed machine learning models, random forest performed best overall in the proposed classifier, with an accuracy of 97.86%, precision of 98%, recall of 77%, F-measure of 86%, and AUC curve of 97%. The most significant determinants of increasing feeding practice were richest household, faculty delivery, use of modern contraception method, the number of children 3-5, women's employment status, maternal age is 25-34, having media exposure, and health-seeking decisions made by mothers were associated positively, whereas not using contraception, home delivery, the total number of children is large, and the sex of the household was male, which was associated negatively with feeding practice during diarrhea in East Africa.
Machine learning (ML) algorithms have provided valuable insights into the complex factors influencing feeding practices during diarrheal disease in under-five children in East Africa. During diarrhea, only 11 of the 100 children received acceptable child feeding practices. More than one-third of the patients received less than usual or nothing. Reducing diarrhea-related child mortality by improving diarrhea management practices is recommended, particularly focusing on the identified aspects.
腹泻是儿童营养不良的主要原因。尽管在腹泻期间进行补液、继续喂养以及在家中增加适当的液体摄入是治疗方案的基石，但在非洲，腹泻疾病期间食物和液体限制却很常见。为了填补方法学和当前证据方面的空白，本研究旨在建立模型并预测决定因素，以增加东非儿童在腹泻暴发期间的喂养行为。
我们使用了2012年至2023年期间从12个东非国家收集的最新人口与健康调查（DHS）统计数据。分析纳入了20,059名5岁儿童的总加权样本。使用Python软件进行数据处理和机器学习模型构建。我们采用了四种机器学习算法，如随机森林（RF）、决策树（DT）、XGB（极端梯度提升）和逻辑回归（LR）。在这项工作中，我们使用准确性、精确性、召回率和AUC曲线等性能评估标准来评估预测模型的性能。
在本研究中，最终分析使用了20,059名5岁儿童。在所提出的机器学习模型中，随机森林在提出 的分类器中总体表现最佳，准确率为97.86%，精确率为98%，召回率为77%，F值为86%，AUC曲线为97%。增加喂养行为的最显著决定因素包括家庭最富有、在医疗机构分娩、使用现代避孕方法、3至5岁儿童数量、妇女就业状况、母亲年龄为25至34岁、有媒体接触以及母亲做出的就医决定呈正相关，而未使用避孕措施、在家分娩、儿童总数多以及家庭性别为男性，这些与东非腹泻期间的喂养行为呈负相关。
机器学习（ML）算法为影响东非五岁以下儿童腹泻疾病期间喂养行为的复杂因素提供了有价值的见解。腹泻期间，100名儿童中只有11名接受了可接受的儿童喂养行为。超过三分之一的患者接受的食物少于平常或没有进食。建议通过改善腹泻管理措施来降低与腹泻相关的儿童死亡率，并特别关注已确定的方面。
