A WebApp Framework for the prediction of e-GFR value and CKD stage using Regression-based Machine Learning Algorithms

Chronic Kidney Disease(CKD) refers to a slow descent in kidney function and has become a serious health concern. Early detection and proper treatment are essential as CKD remains irreversible and unnoticeable until it progresses to advanced stages. Progression of CKD to five different stages depends on Glomeluar Filration Rate(GFR). Hence, timely intervention is crucial for a favorable prognosis and extended lifespan. In this regard, Machine Learning(ML) algorithms have proven to be capable and point in the direction of future disease diagnosis. This paper offers a methodology that uses a range of Regression-based ML algorithms on a real-time CKD dataset with 399 examples to forecast the state of CKD. The work is carried out in two distinct phases. During the initial phase various regression-based ML algorithms like Linear Regression(LR), Lasso Regression, Ridge Regression, K-Nearest Neighbors, Random Forest, and XGBOOST are employed in the CKD dataset to predict the estimated-GFR(e-GFR). In the subsequent phase, the most effective ML algorithm identified from initial phase is intergrated with Flask web framework. This integration enables the prediction of the e-GFR and the CKD stage displayed on the web page. The experimental result showed that the XGBOOST regression method outperformed with an accuracy of 99.6% compared to other regression-based ML algorithms.