Novel Insulin Delivery Profiles for Mixed Meals in Basal-Bolus and Closed-Loop Artificial Pancreas Therapy for Type 1 Diabetes Mellitus

Abstract: Traditional basal-bolus and closed-loop artificial pancreas therapy for type 1 diabetes mellitus were studied in the present work and novel insulin delivery profiles have been identified. Type 1 diabetes is a chronic condition resulting from autoimmune destruction of the pancreatic insulin producing β-cells. Inadequate insulin secretion prevents efficient glucose metabolism and is a serious health risk. Major available treatment modes are multiple daily injections of insulin and insulin pump therapy providing continuous subcutaneous infusion.General insulin regimens for low- and high-fat meals were studied in silico to improve current pump therapy for type 1 diabetes. This involved modifications of the FDA-accepted UVA/Padova metabolic simulation model for evaluations of meals with different absorption rates. Simulations of meals with varied fat content under this modified model demonstrated qualitative replications of published data. Subsequently, an insulin regimen library with optimized regimens under open- and closed-loop settings for a variety of meal compositions was constructed using the particle swarm optimization algorithm. Calculations show that the optimal open-loop insulin delivery profiles for low-fat meals comprise a normal bolus or short square wave depending on the size of the meal. The preferred delivery pattern for large meals is a short insulin wave due to the increased risk for hypoglycemia. Interestingly, the optimal open-loop regimens for high-fat meals are typically biphasic, but can extend to multiple phases for large slow absorbing meals. Furthermore, individual in silico optimizations revealed that patients with high insulin sensitivity could benefit from biphasic insulin deliveries when consuming high-fat meals. Preliminary investigations of the optimal closed-loop regimens under varied fat content also display bi- or triphasic patterns for high-fat meals and are primarily influenced by the carbohydrate content in the meal. The novel insulin delivery profiles identified in this work comprise new and unique waveforms that provide better control of postprandial glucose excursions than existing schemes. Furthermore, the novel regimens are also more or similarly robust to uncertainties in various parameter estimates with the closed-loop schemes displaying superior performance and robustness. The proposed closed-loop strategy does not rely on optimal basal therapy and is therefore a realistic approach that could have real-life applications in an artificial pancreas.