The Automatic Correction Boluses in an Advanced Hybrid Closed-Loop System: Algorithm Performance and Clinical Implications

Abstract

Background:

Automatic correction boluses (ABs) are a key feature of advanced hybrid closed-loop systems, yet their clinical relevance remains incompletely characterized.

Methods:

We retrospectively analyzed insulin delivery and continuous glucose monitoring (CGM) data from 287 individuals with type 1 diabetes using the MiniMed 780G for ≥6 months with ≥80% auto-mode usage, recruited from four diabetes centers in Sicily. We quantified the contribution of ABs to total daily insulin dose (TDD) and total bolus insulin and evaluated associations with demographic factors, pump settings, and dietary habits using multivariate regression and receiver operating characteristic (ROC) analyses.

Results:

ABs accounted for 18.3% of TDD and 33.1% of total bolus insulin. AB/TDD was higher in pediatric participants and inversely associated with diabetes duration and AID use. Active insulin time and carbohydrate intake independently predicted AB/TDD. Higher AB/TDD was associated with lower time in range, higher glucose variability, time above range, and GMI; in pediatric participants, it was also associated with reduced time below range. An AB/TDD ≤20.3% predicted TIR >70% (AUC 0.785) and ≤18.6% predicted GMI <7% (AUC 0.828).

Conclusions:

ABs represent a substantial component of insulin delivery in MiniMed 780G users. Higher AB contribution is associated with poorer glycemic outcomes. Identified thresholds may support system optimization in clinical practice.