City Research Online

Abstract

A wearable artificial pancreas (AP) has been a research goal for over three decades. The aim of this device is to establish effective closed-loop control of blood glucose in patients with type 1 diabetes mellitus (T1DM) using subcutaneous (SC) glucose measurements and SC insulin delivery. The main difficulties that hamper the successful development of a wearable AP concern the stability and accuracy of SC glucose sensing, the predictability of the absorption kinetics of the injected insulin, and finally the performance of the glucose control algorithm. As clinical tests on humans are costly, time consuming, and demand ethical approval, in silico testing of the AP has become a critical feature to facilitate an accelerated development of the AP and, specifically, the control algorithm.

The primary aim of the work reported in this thesis was to explore the use of compartmental modelling techniques with in-built physiological constraints to facilitate the development of a wearable AP.

In particular, the study aimed to extend and evaluate an existing model of whole-body glucose kinetics on a set of data obtained in a clinical trial designed to test the AP algorithm. The model was extended to represent the input-output relationship between the SC insulin and intravenous glucose concentrations. The extended model was re-evaluated in subjects with T1DM under new conditions with the objective to obtain sets of parameters to represent ‘virtual’ subjects with T1DM in the AP simulator. The parameter estimation was completed, but the ‘virtual’ subjects for use in the AP simulator could not be generated due to the uncertain validity of the tested model.

Further objectives included the support for in silico testing of an AP through investigating insulin lispro and interstitial glucose kinetics.

To explore the kinetics of SC administered insulin lispro, ten competing models were proposed assuming a number of physiological effects. The principle of parsimony was used to select the model, which best represented our data. The best model included slow and fast absorption channels, and the presence of local insulin degradation at the injection site.

In order to establish the relationship between the interstitial glucose (IG) and plasma glucose (PG), nine models of IG kinetics were postulated. The model which best represented the experimental data was selected using the principle of parsimony. Two mechanisms explaining the temporal variation in the IG-PG ratio were identified, a zero-order removal of glucose from the interstitial fluid (ISF) and the stimulatory effect of insulin on glucose transfer from the plasma to the ISF. This best model found its use in the simulator to represent SC glucose measurements.

In conclusion, valuable insights were obtained into the mechanisms involved in the insulin and interstitial glucose kinetics, as well as the whole-body glucose kinetics.

Publication Type:	Thesis (Doctoral)
Subjects:	R Medicine > R Medicine (General) R Medicine > RZ Other systems of medicine
Departments:	School of Health & Psychological Sciences School of Health & Psychological Sciences > School of Health & Psychological Sciences Doctoral Theses Doctoral Theses

Export

Downloads