“Algorithmically Smart” Continuous Glucose Sensor Concept for Diabetes Monitoring

In the last 15 years, subcutaneous continuous glucose monitoring (CGM) sensors, portable devices able to measure glycemia almost continuously (1-5 min sampling period) for several days (up to 7), opened new frontiers in the treatment of diabetes, a chronic disease affecting about more than 300 million of people worldwide. However, glucose readings provided by current CGM devices still suffer of accuracy and precision problems. The present work illustrates technical details of algorithms and implementation issues of a new conceptual architecture to deal with those problems and render any commercial CGM sensor algorithmically smarter. In particular, three modules for denoising, enhancement and prediction are assessed on a dataset collected in