A Predictive Algorithm for the Administration of Corrective Insulin Bolus for Decision Support Systems in Type 1 Diabetes

Type 1 diabetes management can be improved by leveraging decision support systems (DSSs), specific tools that, by suggesting therapeutic actions, can assist patients during decision-making process, reducing their daily burden routine. This work proposes a predictive algorithm for DSSs aimed at generating preventive corrective insulin boluses (CIBs) to reduce the duration of hyperglycemic events. Our approach is compared with a recent heuristic-based methodology proposed by Aleppo et al. and it is retrospectively assessed on a dataset recorded in free-living conditions. Preliminary results indicate that the proposed predictive CIB strategy decreases the time above range, largely increases the percentage of time spent in eglycemia without increasing the time spent in hypoglycemia.