Identification of Diabetes Mellitus Through Urine Using Gas Sensor Array and Probabilistic Neural Network Method

Diabetes mellitus is a chronic disease that currently has no cure. It is caused by elevated glucose levels in the body. Invasive examination using blood samples can cause pain and inconvenience for patients, thus prompting the need for non-invasive diagnostic methods, one of which involves using urine samples. In this study, a system was developed to identify diabetes mellitus using semiconductor gas sensors and an Arduino Mega microcontroller to convert analog signals into digital data transmitted to a computer. The Probabilistic Neural Network (PNN) method was used to recognize patterns and classify urine samples as either healthy or diabetic. The experimental results demonstrate that this method can classify urine as healthy or diabetic with an accuracy of 97%.