<![CDATA[Conventionally, examining sugar levels and protein content in the blood is one of them using invasive methods. The method involves medical procedures in the form of injections or procedures related to needles. As a result, if someone has an excessive fear of needles, they are reluctant to carry out an examination. Based on this, we need a system capable of classifying sugar and detecting protein content with non-invasive methods. This system utilizes the physiological conditions of urine and reagents (Benedict and Biuret) for initial diagnosis to make the examination process more effective and efficient. In its implementation, this system uses two main sensors, the TCS230 color sensor and an infrared sensor which extract features from urine physiology in the form of color and detect test tubes. This study uses the K-nearest neighbor method as the classification algorithm and Arduino Mega as a microcontroller. Testing is done by training the system using 105 sample data which includes 70 training data and 15 test data. Testing system accuracy is influenced by the value of K, carried out with scenarios of K = 3, K = 5, and K = 7. The results are system accuracy for each K value of 88.57, 85.71%, and 82.85%. In addition to accuracy, it is known that the average system computing time is 0.734 seconds.]]>
Copyrights © 2022
