Physiology-Based Pharmacokinetic Models in Blood Glucose Control: An Alternative Approach to Tackle Major Hurdles in the development of the Artificial Pancreas
(supported by Bayer)
A core component of Automatic Glycaemic Control is a closed loop control systems that translates glucose measurements into optimal insulin delivery rates. At present, major hurdles in closed-loop concepts for diabetes management are the physiological and pharmacological lag-times as well as the intra- and inter-individual variability of glucose measurement, insulin uptake and response.
To overcome these hurdles several known control theoretical concepts such as Model Predictive Control (MPC) algorithms are combined with mechanistic physiology-based models describing the pharmacokinetics and interactions of glucose, insulin and glucagon.
The main advantages of physiologically-based models are the explicit distinction between substance and patient properties and the possibility to use prior knowledge about individuals to pre-parameterize individualized models and the mechanistic representation of lag-times resulting from lymph and blood flow rates, diffusion and further mass transfer processes occurring within the human body.
Identification or rather the individualization of the PBPK model poses the most challenging task, as the individualized model will be the most critical component of the automatic glucose control system as the performance of control algorithm, and therefore the safety of the patient, will rely on the predictive quality of the model. Thus, the work focuses on the two major workloads, namely, the further development of methods for the characterization of the Mean Distributed Model and the development of a robust and safe online individualization method of the model during the clinical trial or use of the automatic glucose control system in general.