Optimizing Early Diagnosis of Chronic Kidney Disease: A Machine Learning-Based Predictive Model

Chronic Kidney Disease (CKD) is a progressive condition that often remains undetected until its later stages, leading to severe health complications and increased mortality. Therefore, this study presents the implementation of a machine learning-based system for early CKD prediction using the logistic regression algorithm. The study addresses the problem of delayed CKD diagnosis due to vague early symptoms and limited screening efficiency in traditional clinical workflows. The development of the system utilizes a clinical dataset from the UCI Machine Learning Repository made up of 400 patient records with 24 features, including demographic, clinical and laboratory parameters. Data preprocessing techniques were employed including label encoding, imputation of missing values and feature scaling for improving the quality of the data. Then, feature selection was conducted using the SelectKBest method with mutual information to identify the most relevant predictors. The logistic regression model was trained and evaluated using an 80:20 data split and the result of the implementation presents that the model achieved an accuracy of 97.5%, precision of 96.7%, recall of 98.2%, F1-score of 97.4% and a ROC-AUC score of 0.99. This work underscores the effectiveness of logistic regression in medical diagnostics and highlights the value of machine learning in facilitating early detection and timely treatment of CKD.