20 Juillet – Thesis defense - Loïc Olçomendy
14 h Amphi D - ENSEIRB-MATMECA (Talence)
In silico development of a bioelectronic islet-based artificial pancreas.
The development of new therapeutic pathways for the treatment of diabetes requires highly interdisciplinary research. For the last fifty years, the need for interdisciplinarity in translational research projects has been further strengthened by the digital revolution. The Artificial Pancreas (AP) is a prime example of medical device developed thanks to the contribution of scientists, engineers, and mathematicians. APs, as “all-in-one” diabetes management systems now appear as a standard of care to restore the glucose homeostasis of type 1 diabetic patients. These semi-automated closed-loop devices successfully replace the defective endogenous insulin secretion by the continuous infusion of finely-tuned exogenous insulin boluses.
Our research consortium developed a biosensor enabling the real-time characterisation of pancreatic islet algorithms via non-invasive electrophysiological measurements. We hypothesize that, in contact with T1D patient interstitial fluids, healthy islets embedded in this wearable biosensor could provide an indication on the patient’s need in insulin and thus constitute a valuable physiological input for the AP. This thesis work investigates the contribution of numerical simulation to the development of an AP system involving this innovative sensor.
The introductory chapter of the manuscript provides the scientific context of this work, which lies at the intersection of biology, electrophysiology, electronics, control theory and diabetology. The second chapter then presents the necessary material and the methods developed to achieve the results described and discussed thereafter.
Our research approach was divided into two separate simulation pathways. A first pathway, described in chapter 3, intended to validate the biosensor’s working principle by exploiting the advantages of numerical simulation. This approach is however not realistic from a clinical standpoint as it uses intravenous routes. In particular, we achieved excellent glucose control using a regulation scheme based on electrically-characterised endogenous islets algorithms. Integrating the islet models in an AP architecture, we then developed a second simulation pathway to assess the potential contribution of our biosensor to type 1 diabetes treatment. This pathway, described in chapter 4, uses a more clinically realistic configuration of the virtual patient simulator which enables a comparison between our biosensor-based AP and standard treatment approaches. These preliminary results are promising: the biosensor-based AP permitted a satisfactory glucose control, even in challenging conditions (meals containing high glycaemic loads). The identification of the benefits and limitations of our simulation campaign gives rise to a discussion on the contribution of numerical simulation to the development of new solutions for the treatment of diabetes. As a conclusion, we define general guidelines in an attempt to lay the groundwork for a future real-world implementation of a biosensor-based AP system.