DefinePK

Abstract

Childhood Anemia has always been a great concern in Pakistan, as the issue is directly linked with the adverse-growth, development, and cognitive out- comes. To provide prompt intervention it is necessary to identify potentially vulnerable children earlier, but access to the classic approaches of diagnosis is frequently impeded by its reduced availability and efficiency. The article implements the methodology/strategy of exploratory data analysis (EDA) and machine learning (ML) to come up with an efficient predictive framework of childhood anemia. EDA has been applied to nationally representative health survey data to predict major predictors in the health survey, examine relations among the variables, and discover patterns of data. A variety of ML models were trained such as logistic regression, decision tree model, random forest model, k-nearest neighbor model, and gradient boosting model evaluated and their predictive power was measured with model optimization used to improve accuracy and reliability. The findings show that the ML models are more efficient than the traditional cases in the prediction of the anemia status, and relatively, child age, household wealth,
maternal education, and nutritional indicators are shown to be the most suc- cessful predictors. This paper illustrates how EDA is complemented by ML can help realize data-driven health interventions and outlines a cost-effective and scalable solution to measure anemia risk in resource-constrained envi- ronments, such as Pakistan.
Data Availability:
The data used in this study is extracted from the Pakistan Demographic Health Survey 2017-18 [PDHS].
Funding:
The authors received no specific funding for this work.
Competing Interests:
The authors have declared that no competing interests exist.
List of Abbreviations:
ML, Machine Learning; Anemia; EDA, Exploratory Data Analysis; SVM, Support Vector Machine; LR, Logistic Regression; DT, Decision Tree; RF, Random Forest; GB, Gradient Boost; NB, Naive Bayes; AI, Artificial Intelligence; PDHS, Pakistan Demographic Health Survey; ROC, Receiver Operating Characteristic; AUROC, Area Under Receiver Operating Characteristic