Mealtime correction insulin advisor for CGM-informed insulin pen therapy

With improvements in the accuracy and reliability of continuous glucose monitoring (CGM), the stage is set for new algorithmic approaches to the treatment of Type 1 diabetes. While recent efforts to build a closed-loop artificial pancreas device are encouraging, the artificial pancreas will not apply to patients who prefer to inject insulin manually or who simply will not submit to fully automatic closed-loop control. Therefore, it is of interest to see whether non-pump users can benefit from algorithmic insulin advisory systems. In this paper we present a mealtime correction bolus advisor for patients on “multiple daily injection” (MDI) therapy, taking advantage of the ability to estimate the patient's metabolic state in real time using both CGM and manual reports of insulin delivery. The state estimation process for this is informed by knowledge of the patient's daily injection of long-acting insulin through the novel concept of a virtual basal rate profile. Preliminary in silico trials indicate that the advisor can result in significantly improved control (reduction of up to 1% saturation of hemoglobin A1C) for patients currently struggling with sustained hyperglycemia.