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

Background/Objectives: Early childhood anemia is a severe public health concern and the most common blood disorder worldwide, especially in emerging countries. This study examines the sources of childhood anemia in Ghana through various societal, parental, and child characteristics. Methods: This research used data from the 2022 Ghana Demographic and Health Survey (GDHS-2022), which comprised 9353 children. Using STATA 13 and R 4.4.2 software, we analyzed maternal, social, and child factors using a model-building procedure, logistic regression analysis, and machine learning (ML) algorithms. The analyses comprised machine learning methods including decision trees, K-nearest neighbor (KNN), logistic regression, and random forest (RF). We used discrimination and calibration parameters to evaluate the performance of each machine learning algorithm. Results: Key predictors of childhood anemia are the father's education, socioeconomic status, iron intake during pregnancy, the mother's education, and the baby's postnatal checkup within two months. With accuracy (94.74%), sensitivity (82.5%), specificity (50.78%), and AUC (86.62%), the random forest model was proven to be the most effective machine learning predictive model. The logistic regression model appeared second with accuracy (67.35%), sensitivity (76.16%), specificity (56.05%), and AUC (72.47%). Conclusions: Machine learning can accurately predict childhood anemia based on child and paternal characteristics. Focused interventions to enhance maternal health, parental education, and family economic status could reduce the prevalence of early childhood anemia and improve long-term pediatric health in Ghana. Early intervention and identifying high-risk youngsters may be made easier with the application of machine learning techniques, which will eventually lead to a healthier generation in the future.