Background: MDI therapy requiring carbohydrate (carb) counting can be burdensome for persons with diabetes (PWD). We conducted an in-silico feasibility study assessing the glycemic performance of a post-prandial rather than a carb-based bolus optimizer.

Methods: The data were collected from seven adults with T1D (Gender:M2/F5; Age:22-45) on MDI using the Guardian™ Connect continuous glucose monitoring system. All participants had three bolus events matched to three meals for ≥7 days. In the training stage, a glucose-insulin minimal model extracted the rate of glucose appearance in blood (Ra) for each meal. The Ra was taken as an input to adjust in-silico simulated bolus amounts until the greatest time in sensor glucose range (70-180 mg/dL, TIR) within 4hrs was reached. In the testing stage, two bolus strategies were tested: the average of the adjusted bolus from the training set (Uniform Bolus) and the average of the adjusted bolus from the training set associated with the time of day (TOD) of the meal announcement (TOD Bolus).

Results: We found increased TIR for TOD Bolus strategy compared to the User Bolus (baseline) strategy (See Table below).

Conclusion: The data suggested that the automated bolus optimizer without carb counting may allow for an increased time spent in the target range, without comprising time spent below range while reducing the burden of carb counting for PWD on MDI.


B. Jiang: Employee; Self; Medtronic. Y. Zhong: Employee; Self; Medtronic. P. Agrawal: None. T.L. Cordero: Employee; Self; Medtronic. R. Vigersky: Employee; Self; Medtronic.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at