Diabetes is a highly extended disease worldwide, associated with vascular complications or strokes. Besides glucose monitoring is absolutely necessary for the prevention of the different adverse effects related with that disease, nowadays the most popular techniques to measure the glucose levels are invasive. Thus, there is a need of innovation in pursuit of find minimally invasive techniques for glucose measuring.
The goal of this use case is to develop an approach for a prototype for glucose monitoring that will be able to transform high frequency signal into glucose level using novel neuromorphic algorithms, such as Spiking Neural Network (SNN). This kind of algorithms offers strong advantages in the prediction, speed, and energy efficiency. Thus, it is also desired to compare the results of the SNN with other traditional algorithms, such as ANN.
Therefore, this use case is focused on glucose monitoring, evaluating the developed neuromorphic hardware in context of wearables and portable healthcare. For that purpose, high frequency sensors will be used to build a setup, which will be used for data gathering by measuring different dilutions of glucose and distilled water in cuvettes. Once the database is created, the SNNs algorithm will be implemented to detect the glucose level. The problem will be addressed in terms of classification, with several predictable glucose levels that will vary between 2 and 5 classes. The obtained model architecture will be integrated into the developed neuromorphic hardware.
EESY and IFAG collaborate for the development of a minimally invasive glucose monitoring setup by complementing neuromorphic algorithms and high frequency sensors.