Related Experiment Video
Updated: Jul 23, 2026
12:18
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
8.2K
Leveraging machine learning algorithm to predict minimum dietary diversity among children aged 6-23 months in
Naol Gonfa Serbessa1, Siraj Muhidin Degefa1, Beriso Alemu Hailu1
1Department of Health Informatics, College of Health Science, Mattu University, Mattu, Ethiopia.
PLOS Global Public Health
|February 26, 2026
Summary
Machine learning accurately predicts minimum dietary diversity in Ethiopian children. Key factors include delivery location, household head
Area of Science:
- Child Nutrition
- Public Health
- Machine Learning Applications
Background:
- Inadequate nutrient-rich food intake impacts child development, potentially causing delays and disorders.
- Limited evidence exists on predictors of dietary diversity in children.
- Minimum Dietary Diversity (MDD) remains a significant public health concern in Ethiopia, with notable regional and socioeconomic disparities.
Purpose of the Study:
- To train and evaluate eight machine learning algorithms for predicting minimum dietary diversity in Ethiopian children aged 6-23 months.
- To identify key predictors of minimum dietary diversity using machine learning and Shapley Additive Explanations (SHAP).
Main Methods:
- Utilized secondary data from the Ethiopian Demographic and Health Survey (EDHS) spanning 2005-2019 (n=8,996 children aged 6-23 months).
- Employed STATA 17 for data extraction and Python 3.11 for data cleaning, coding, and analysis.
- Tested eight machine learning algorithms: logistic regression, random forest, K-Nearest Neighbors (KNN), Multilayer Perceptron (MLP), Support Vector Machine, Naive Bayes, Extreme Gradient Boost (XGBoost), and AdaBoost.
Main Results:
- The Random Forest classifier achieved the highest performance (Accuracy=82%, AUC=89%) in predicting minimum dietary diversity.
- Key predictors identified by the Random Forest model and SHAP analysis include place of delivery, household head's sex, water source, place of residence, child's age, number of children under five, mother's age, and household size.
- The study highlights significant regional and socioeconomic inequalities affecting minimum dietary diversity.
Conclusions:
- Machine learning, particularly the Random Forest model, is effective in predicting minimum dietary diversity among young children in Ethiopia.
- Identifying at-risk populations through machine learning can inform targeted nutrition interventions.
- Addressing socioeconomic and regional disparities is crucial for improving child nutrition outcomes.
Related Concept Videos
01:23Pharmacokinetics in Pediatric Patients: Overview and Drug Absorption
Understanding the physiological differences in the pediatric population is crucial for effective pharmacotherapy. Neonates, infants, and children exhibit significant variations in gastric pH, gastric emptying time, intestinal transit time, and biliary function. These variations profoundly affect oral drug absorption, necessitating a nuanced approach to pediatric dosing.Neonates present with a unique physiological profile, having a gastric pH greater than 4 and faster and more irregular gastric...
01:24Pharmacokinetics in Pediatric Patients: Drug Metabolism
In pediatric care, understanding the nuances of hepatic drug metabolism is crucial, as it significantly differs from that of adults. This divergence is primarily due to the developmental stage of drug-metabolizing enzymes, which affects how medications are processed in the body. In neonates, for instance, the activity of Phase I enzymes—critical for the initial breakdown of drugs—is markedly reduced, functioning at just 20–40% of the levels seen in adults. This reduction poses a challenge in...