Explainable deep learning for chronic kidney disease prediction

Abstract

Chronic Kidney Disease (CKD) is a progressive and incurable condition affecting over 10% of the global population (Kovesdy, 2022). The disease imposes a significant financial burden on healthcare systems due to the high costs of medications, dialysis, and specialized treatments. Early detection and continuous monitoring are essential in mitigating CKD's effects, but traditional diagnostic methods, relying on clinical tests and expert analysis, can be timeconsuming and subjective. This research aims to develop an explainable deep-learning model to predict CKD progression by analyzing historical patient data. Unlike conventional black-box models, this approach will incorporate explainability techniques such as Local Interpretable Modelagnostic Explanations (LIME) and (SHapley Additive exPlanation) SHAP to provide transparent insights into risk factors, enabling medical professionals to make data-driven decisions.