Estudo de processo para a análise de glicose sanguínea

Abstract

Diabetes is one of the major endocrine diseases that poses a serious threat to health. However, the current method, which measures the glucose level from blood sampling, is too unpleasant and painful to use often. Non-invasive glucose monitoring is the most attractive approach to diabetes management, allowing for more frequent and continuous measurement without pain and bleeding. The implementation of a non-invasive equipment through infrared, can enable monitoring in a less uncomfortable way and that can be performed at any time without the need to perform a puncture. This, in turn, can impact on minimizing the lack of glycemic monitoring that is often avoided, precisely because it is uncomfortable for the patient. Photoplethysmography (FPG) is a non-invasive optical technique that has been used to monitor changes in glycemic indices. For the mathematical resolution of the glucose estimate, the MATLAB software was used. Thus, a prototype capable of estimating glucose was implemented non-invasively, using two regression algorithms, Decision Tree Regression and Gaussian Exponential Regression. In in vitro tests, the algorithms were able to predict glucose, unlike in vivo tests, where further research is needed.