Advanced Artificial Pancreas Systems to Enable Fully Automated Glycemic Control in Type 1 Diabetes Mellitus

PROJECT ABSTRACT Despite advancements in automated insulin delivery systems that demonstrate superior glycemic outcomes, challenging clinical issues remain with the continued need for accurate carbohydrate counting and meal announcement combined with slow-acting subcutaneous insulin that contribute to repeated occurrences of postprandial hyperglycemia. This clinical gap in glycemic control is addressed by the proposed project. We intend to leverage our existing advancements in pattern recognition and anticipation to refine and test a fully automated closed-loop system that does not require meal announcements. This system is unique in that it relies on pattern recognition, particularly surrounding meals, in an anticipatory manner so that insulin action is taken in advance of an expected meal rather than the predominantly reactive solutions that are often proposed. We intend to further test and tune this system within the context of accelerated insulins to further improve glycemic outcomes and remove barriers to use of these systems by lowering burden of disease management. We have already developed the foundational tools needed for a fully automated closed-loop and will plan for further refinement by leveraging data from our accumulated clinical studies database. We plan to conduct 2 pilots studies followed by a larger, 7-month main study to evaluate the following specific aims of this proposal: 1) We hypothesize that novel faster acting insulin analogs can be safely used to provide more effective fully automated closed-loop control. Our first pilot will test the differences between Fiasp and a novel formulation of aspart (AT247) in a supervised trial. This trial results will inform the selection of the insulin used in the subsequent trials. 2) We hypothesize that, with a design adapted to these new insulin formulations, fully automated closed-loop control is a valid clinical alternative to hybrid closed-loop systems, the current state- of-the-art systems. 3) Lastly, we hypothesize that further adding glycemic disturbance anticipation (e.g. generally surrounding meals) to this fully automated closed-loop control system will further improve glycemic outcomes compared with a fully automated closed-loop system without disturbance anticipation.