Early Childhood Anemia in Ghana: Prevalence and Predictors Using Machine Learning Techniques

Area of Science:

• Public Health
• Pediatrics
• Computational Biology

Background:

• Early childhood anemia is a significant global health issue, particularly in developing nations.
• It represents the most prevalent blood disorder worldwide, impacting vulnerable populations.
• Understanding its determinants in specific contexts, like Ghana, is crucial for effective public health strategies.

Purpose of the Study:

• To investigate the multifaceted determinants of childhood anemia in Ghana.
• To identify key societal, parental, and child-specific factors contributing to anemia prevalence.
• To evaluate the efficacy of machine learning models in predicting childhood anemia.

Main Methods:

• Utilized data from the 2022 Ghana Demographic and Health Survey (GDHS-2022) involving 9353 children.
• Employed logistic regression, decision trees, K-nearest neighbor (KNN), and random forest (RF) algorithms for analysis.
• Assessed machine learning model performance using discrimination and calibration parameters.

Main Results:

• Significant predictors of childhood anemia include father's education, socioeconomic status, maternal iron intake during pregnancy, mother's education, and postnatal checkups.
• The random forest model demonstrated superior predictive performance with 94.74% accuracy and 86.62% AUC.
• Logistic regression showed moderate predictive capability with 67.35% accuracy and 72.47% AUC.

Conclusions:

• Machine learning models can accurately predict childhood anemia using child and paternal characteristics.
• Interventions targeting maternal health, parental education, and socioeconomic status are vital for anemia reduction.
• ML techniques can facilitate early identification of high-risk children, promoting healthier future generations.