Development of Detection and Classification Techniques with Application to Non invasive Chronic Kidney Disease Monitoring

Abstract

This research aims to explore the feasibility of salivary analysis for Chronic Kidney Disease (CKD) detection and thereby design a non-invasive automated mechanism to detect CKD through analysis of human saliva samples. Kidney disease is commonly identified through a blood-based screening process. Although the blood-based screening process is accurate, it has many drawbacks due to its invasive sample collection approach. Therefore, there is a need for a reliable non-invasive prediction system. New research findings reveal that saliva-based testing can be used for detecting kidney disease. Saliva-based analysis has several advantages, and the most significant advantage is that it offers a non-invasive way of detection. In this work, we have developed deep learning Convolutional Neural Networks (CNN) and a novel diagnosis method to detect CKD non-invasively. We have examined the concentration of urea in the saliva sample to detect the disease. Significantly higher levels of salivary urea are found in patients with kidney disorders. The proposed learning networks are trained and tested with the hardware sensing module. As the Support Vector Machine (SVM) classifier is more effective in classifying the samples, we have integrated this classifier with the convolutional networks. The results of this study show that the proposed learning models can classify the samples with the lowest possible error of misclassification. The deep learning algorithms introduced in this work are found to significantly reduce the limitations associated with using traditional methods and further improves the classification performance. The detection module and classification algorithms substantially advance the current methodologies, and it provides more accurate predictions compared to conventional methods