Predictive Modeling to Identify Stunting Risk in Children

Area of Science:

• Public Health
• Pediatrics
• Data Science

Background:

• Indonesia continues to face a significant burden of childhood stunting, with profound short- and long-term health and economic consequences.
• Stunting impacts include increased morbidity, mortality, impaired growth, higher chronic disease risk, and reduced future productivity.

Purpose of the Study:

• To identify the primary risk factors associated with stunting in Indonesian children.
• To develop and evaluate a predictive model for identifying children at risk of stunting.

Main Methods:

• Analysis of the 2018 Indonesian Basic Health Research database, including 13,106 children under 5 and their mothers.
• Bivariate analysis to identify significant risk factors, followed by a decision tree model for prediction.
• Receiver Operating Characteristic (ROC) curve analysis to evaluate model performance.

Main Results:

• The national stunting rate was 25.8%.
• Key predictors identified by the decision tree model include age, sex, birth weight, birth length, maternal education, handwashing habits, and exclusive breastfeeding.
• The predictive model achieved 73.8% accuracy, with an Area Under the Curve (AUC) of 63.7% on the ROC curve.

Conclusions:

• The developed prediction model demonstrates acceptable accuracy in assessing stunting risk.
• The decision tree model effectively differentiates between stunted and non-stunted children across various age groups, as supported by ROC curve analysis.