Prediction of anemia and associated risk factors among under five years children in Rwanda using machine learning techniques

Abstract

The health problem of anemia affects both developed and developing nations, having a significant negative impact on both social and economic development and human health. Rwanda Demographic and Health Survey (RDHS) 2014–2015 reported that 37% of children in Rwanda between the ages of 6 to 59 months are anemic. The study aimed at identifying the factors affecting anemia in under five years Rwandan children and predicting its status using machine learning techniques. In this research, secondary data from the RDHS 2019–20 were used. The risk variables linked to anemia in under five years children were determined with the help of bivariate and multivariate analyses. Machine learning models including K-Nearest Neighbor (KNN), Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM), and Decision Tree (DT) were used to predict its status. Several evaluation metrics including Area Under the Curve (AUC), Confusion Matrix, Recall, Accuracy, F1 Score, and Precision were used to assess the performance of the created models. According to the multivariate analysis findings, the factors for anemia in under five years Rwandan children included wealth index, child age, the mother's anemic status, and having recently had diarrhea. In predicting using ML models, the evaluation metrics used show the highest results for the Random Forest model compared to others. The results showed that Random Forest had 70% precision, 70% recall, 70% f1 score, and 70% AUC, indicating that it produced 70% accurate outcomes. The study concludes that Machine learning is a powerful tool that is being used in health issues for predicting diseases and identifying hidden patterns which couldn’t be revealed easily by other traditional methods.