Correlation-Based Analysis of Biomarkers for Predicting Chronic Kidney Disease

Abstract: Chronic Kidney Disease (CKD) presents a considerable public health challenge, often detected at advanced stages when intervention is less effective. This study conducts a correlation-based analysis of essential biomarkers for predicting Chronic Kidney Disease (CKD), focusing on indicators like serum creatinine, blood urea, albumin, haemoglobin and blood pressure. Utilizing correlation matrix analysis, the study identifies positive and negative correlations among these biomarkers, revealing key associations with CKD progression. The correlation analysis revealed strong positive relationships between CKD stages and biomarkers such as haemoglobin (0.77) and specific gravity (0.73), both critical for assessing disease progression. Conversely, negative correlations, such as between serum creatinine and sodium (-0.69) and albumin and CKD class (-0.63), highlight electrolyte imbalances and kidney damage markers that commonly manifest in advanced CKD. The findings highlight biomarkers with high predictive value, contributing to enhanced early detection and risk assessment. These findings underscore the predictive value of key biomarkers, providing insights for refining machine learning models and enhancing CKD diagnosis and management strategies for early intervention and personalized treatment. This analysis supports the refinement of machine learning models and aids in developing more effective CKD diagnosis and management strategies.

Keywords: Chronic Kidney Disease (CKD), Machine Learning, Biomarkers.