A Systematic Review of Chronic Kidney Disease Prediction in Nigeria using Machine Learning Techniques

Abstract

Chronic kidney disease (CKD) is a major public health problem in Nigeria, where delayed presentation, limited diagnostic capacity and low awareness contribute to under-diagnosis. Early identification of individuals at risk may help delay progression to end-stage renal disease. Machine learning (ML) provides an opportunity to improve risk stratification using routinely collected clinical data. This systematic review synthesized studies that used Nigerian clinical data to develop supervised ML models for CKD prediction or diagnosis. A systematic search of PubMed, Scopus, IEEE Xplore, Google Scholar and African Journals Online was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines for studies published between January 2010 and April 2025. Methodological quality and risk of bias were assessed using the Prediction Model Risk of Bias Assessment Tool (PROBAST). Six empirical studies met the eligibility criteria. Reported model accuracy ranged from 81.3% to 99.4%, with Random Forest, Gradient Boosting, Deep Neural Networks and hybrid approaches among the methods used. Serum creatinine, age, blood pressure, blood urea nitrogen and haemoglobin were frequently reported predictors. Major limitations included small and heterogeneous datasets, limited interpretability, inadequate external validation and limited translation of models into routine clinical practice. Future research should prioritize multicentre prospective datasets, transparent reporting, external validation and explainable ML approaches to support clinically useful and equitable CKD prediction in Nigeria.