Predictive glucose alarms (which alert the patient when their blood glucose is about to fall dangerously low or rise dangerously high) and model-based algorithms for smart glucose control are only possible if diabetics can properly predict future blood glucose levels. Based on evidence from clinical studies, control-oriented graybox process models can best manage this task. As part of the current study, researchers are looking into how outpatient data from MDI patients may be used to reliably parametrize such models. Using a preprocessing technique, it is possible to identify (i.e., locate) data segments that are complete and sensible enough to be used for system identification. For the current study, we are focusing on the prediction of postprandial glucose trajectories, particularly those made at the time of meal consumption. Model-based insulin dosage optimization necessitates accomplishing this difficult but critical task. Predicting such postprandial glucose excursions can be done using the identified process models, which have been shown to work.

.