The impact of insulin pharmacokinetics and pharmacodynamics on the closed-loop artificial pancreas

Abstract

This paper investigates the expected clinical performance and robustness envelope (e.g., postprandial blood glucose peak and low blood glucose condition risk) of an artificial pancreas (AP) with respect to the pharmacokinetic and pharmacodynamic (PK/PD) characteristics of different insulin delivery options (i.e., combinations of insulin formulations and routes). Specifically, an insulin PK/PD based design method that enables the investigation of the clinical performance and robustness envelope using the internal model control methodology is introduced. The controller based on the proposed method was evaluated on the Universities of Virginia and Padova FDA-accepted metabolic simulator. The robustness analyses suggest that the controller can achieve safe glucose regulation over a wide range of insulin PK/PD characteristics. The simulation results suggest that the controller with only two parameters (the pharmacokinetic time constant and the controller time constant) can achieve satisfactory performance (meal disturbance rejection within 3.5 h) over the PK/PD range, demonstrating that a faster acting insulin delivery option achieves superior meal disturbance rejection without inducing a dangerous overshoot (i.e., low blood glucose condition). Additionally, two sample clinical results using fast acting insulin delivery options (inhaled and intraperitoneal) are given to illustrate the benefit of improving PK/PD characteristics.