Abstract
A large percentage of people globally suffer from chronic kidney disease (CKD), a serious health concern. Effective diagnosis, treatment, and referral of CKD depend heavily on early identification and prediction of the disease. However, it is difficult to evaluate and derive significant insights from health data due to its vast and complicated nature. Engineers and medical researchers are using data mining techniques and machine learning algorithms to create predictive models for chronic kidney disease (CKD) in an effort to address this issue. The goal of this research is to create and validate predictive models for chronic kidney disease (CKD) based on a variety of clinical factors, including albuminuria, age, diet, eGFR, and pre-existing medical problems. The objective is to estimate the likelihood of renal failure, which may necessitate kidney dialysis or a transplant, and to evaluate the degree of kidney disease. With the use of this knowledge, patients and healthcare providers should be able to make well-informed decisions about diagnosis, treatment, and lifestyle changes. Patterns in the gathered data can be found, and future incidence of CKD or other related diseases can be predicted, by utilising MLT such as ANN and data mining techniques. Finding novel characteristics linked to the onset of renal disease and adding more trustworthy data from CKD patients. The best algorithm to categorise the data as CKD or NOT_CKD is chosen throughout the design process, and the data is then classified according to this differentiation. Estimated glomerular filtration rate (eGFR), which offers important details about the patient's current kidney function, is used to classify cases of chronic kidney disease. By combining complete patient data with machine learning algorithms, this research advances the diagnosis of chronic kidney disease (CKD) and improves patient outcomes.