Persistent Kidney Disorder may seem to be knocking at the door of every community; it carries along its nature of morbidity and mortality along with it and various issues leading to the deterioration of health. Detection is rarely easy due to the asymptomatic presentations at early stages. With luck, early diagnosis of CKD allows timely intervention to slow the disease down. Deep learning models could really help clinicians monitor such conditions since they can rapidly and accurately spot such conditions. This paper elaborates on the use of machine learning in the diagnosis of CKD. The dataset is retrieved from the deep learning repository of the University of California, Irvine (UCI).  The framework aims at patients with CKD diagnosed as a result of the disease and examines whether the patients need to be treated. Various deep learning engines such as CNN, MobileNet, VGG16 were trained based on the sufficient models for kidney diagnostics. Among these, random forest gives the best of all accuracies. An integrated model proposed by the evaluation of errors of these models combined logistic regression with random forests using a perceptron for enhanced accuracy. This approach can foster the possible application of more complex clinical data for effective disease diagnosis.