Glucagon Administration Strategies for a Dual-Hormone Artificial Pancreas

V. Batora, D. Boiroux, M. Hagdrup, M. Tarnik, J. Murgas, S. Schmidt, K. Nørgaaard, N.K. Poulsen, H. Madsen, J.B. Jørgensen

Abstract

In this paper, we describe and evaluate Model Predictive Control (MPC) algorithms for an Artificial Pancreas (AP) with both insulin and glucagon. Such a device is called a dual-hormone AP. We evaluate the performance of two different glucagon administration strategies and compare them to a strategy using insulin only that serves as a reference strategy. In all cases, the dosing of the insulin is based on the same MPC algorithm. Glucagon is included as a safety feature in the closed-loop system, and the administration of insulin is done in a non-aggressive way such that it does not anticipate glucagon administration. Insulin and glucagon are never administered simultaneously; the switch between insulin and glucagon administration is based on either a measured glucose level threshold with hysteresis or a prediction of future hypoglycemia. We simulate cases where meals are correctly bolused and not bolused, as well as cases with insulin sensitivity changes. The results indicate that both switching strategies for a dual-hormone AP reduce the time spent in hypoglycemia significantly. Moreover, the glucagon control strategy using glucose predictions in the switch allows earlier glucagon intervention and further reduces the time spent in hypoglycemia without overdosing glucagon.

Downloads

• Submitted manuscript

Reference

V. Batora, D. Boiroux, M. Hagdrup, M. Tarnik, J. Murgas, S. Schmidt, K. Nørgaaard, N.K. Poulsen, H. Madsen, J.B. Jørgensen: “Glucagon Administration Strategies for a Dual-Hormone Artificial Pancreas”, submitted to IEEE Transactions on Biomedical Engineering, June, 2015