Design of predictive models for identifying the early stage chronic kidney disease using machine learning techniques

Abstract

Chronic Kidney Disease (CKD) is a significant public health newlineconcern, with a rising mortality rate and incredibly high costs associated with newlinedialysis and transplantation. Early-stage CKD warning signs are not obvious. newlineThe main difficulty in managing kidney disease is that many people are newlineunaware of their condition until significant damage has occurred. Delay in newlinedetecting kidney malfunction has been linked to an increased risk of newlineprogression to End-Stage Renal Disease (ESRD). The risk of CKD newlineprogression can be reduced if the CKD is diagnosed at its early stage. This newlinechallenge inspires the concept of automated early-stage CKD prediction, newlinewhich would be extremely beneficial in terms of patient healthcare and cost. newlineHealth records of the patients can be used for automated disease diagnosis to newlinesupport clinical decision-making. In disease diagnosis, Artificial Intelligence newline(AI) provides more erudite and expert services. Machine Learning (ML) newlinealgorithms are widely applied in healthcare to predict and classify diseases. newlineThe focus of this research is to use ML techniques to create predictive models newlinefor detecting early-stage CKD. Support Vector Machine-CKD (SVM-CKD), newlineDeep Neural Network-CKD (DNN-CKD), and Multi-Layer Perceptron- newlineSynthetic Minority Oversampling Technique-CKD (MLP-SMOTE-CKD) newlinemodels are proposed to predict the early-stage CKD. Features from blood tests, urine tests, and patient history may be included in medical databases collected for any disease diagnosis. ML techniques applied to a medical database may aid in identifying features that can be used as early-stage CKD predictors. Medical data obtained from a variety of sources are diverse. It may include features with numerical and nominal values. Selecting risky factors with the same feature selection Renal function deterioration is linked to Cardiovascular Disease (CVD). Patients with CVD, on the other hand, are frequently underdiagnosed and undertreated for CKD because clinical diagnosis and treatment are often focused on a single organ in the early stages. The MLP-SMOTE-CKD prediction model is built to predict CKD in patients with CVD or at high risk of CVD and to study the comorbidity of CKD in CVD patients. Healthcare data is frequently imbalanced newline newline