Statistical Analyses for Key Risk Factor Identification and Prediction of Chronic Kidney Disease

Chronic renal disease, also known as chronic kidney disease (CKD) is a common disease and is a concern of public health management. Effective techniques for early CKD prediction are desirable. Given a set of biomarkers, Machine learning techniques are known for predicting CKD. This work aims of predicting CKD given clinical data. The proposed work suggests a methodology that includes data prepossessing (i.e. data cleaning, addressing null values and normalizing), applying statistical methodologies for finding key risk factors. Finally, using the most significant risk factors, machine learning techniques is applied to prognosticate the onset of CKD. The proposed approach has been tested with two data sets and proves to be fast, cost-effective and accurate compared to the existing state of the art techniques.Clinical relevance- Prognosticating CKD with a higher accuracy using a minimum number of risk factors is a significant aspects of healthcare informatics, where the treatment cost and predictive results both are optimized towards the betterment of patients.