ENZYMATIC UREA BIOSENSOR WITH OPTICAL TRANSDUCTION FOR RENAL DISEASE MONITORING USING CONVOLUTIONAL NEURAL NETWORK

This study describes the development of an enzymatic biosensor for urea detection with optical transduction, designed to monitor Chronic Kidney Disease (CKD). CKD is a serious condition characterized by the progressive and irreversible loss of kidney function, compromising the body's ability to maintain homeostasis. The primary goal of the developed biosensor is to assist in the early diagnosis of CKD, which is essential for improving patient prognosis. The biosensor's production involved the immobilization of the urease enzyme on a polyvinylpyrrolidone film, using Nessler's reagent for optical transduction. Characterization of the biosensor was performed through enzymatic kinetic activity analyses, FTIR spectroscopy, and pH and UV-Vis measurements. Urea levels were indirectly detected using digital image classification techniques with Convolutional Neural Networks (CNNs). The Mobile Nets model, used for training and classifying the images, achieved an accuracy of 93.0%, correctly identifying 139 out of 150 tested samples. The results highlight the potential of this biosensor as an innovative, cost-effective, and non-invasive tool with high sensitivity for the diagnosis and monitoring of CKD. It represents an innovative and effective approach to advancing kidney health monitoring.