Prediction of Stunting Among Under-5 Children in Rwanda Using Machine Learning Techniques.

Area of Science:

• Public Health
• Machine Learning Applications
• Pediatric Nutrition

Background:

• Childhood stunting remains a significant public health concern in Rwanda, with a 33% rate in 2020.
• Global child deaths due to malnutrition highlight the urgent need for effective early detection and treatment strategies.

Purpose of the Study:

• To develop and evaluate a machine learning model for predicting stunting in Rwandan children.
• To identify key predictors of stunting for targeted interventions.

Main Methods:

• Utilized secondary data from the Rwanda Demographic and Health Survey 2019-2020.
• Employed stratified 10-fold cross-validation and various machine learning classifiers.
• Selected the best performing model based on accuracy, sensitivity, specificity, and F1 score.

Main Results:

• A gradient boosting classifier achieved 80.49% training accuracy and 79.33% test accuracy.
• The model demonstrated high specificity (94.49%) in identifying non-stunted children.
• Key predictors identified include maternal height, child's age, mother's education, and birth weight.

Conclusions:

• Machine learning offers a viable approach for accurate early detection of childhood stunting in Rwanda.
• The developed model can inform public health policies for stunting prevention and control in children under five.