In Silico Preclinical Trials: Methodology and Engineering Guide to Closed-Loop Control in Type 1 Diabetes Mellitus

Abstract

This article sets forth guidelines for in silico (simulation-based) proof-of-concept testing of artificial pancreas control algorithms. The goal was to design a test procedure that can facilitate regulatory approval [e.g., Food and Drug Administration Investigational Device Exemption] for General Clinical Research Center experiments without preliminary testing on animals. The methodology is designed around a software package, based on a recent meal simulation model of the glucose-insulin system. Putting a premium on generality, this document starts by specifying a generic, rather abstract, meta-algorithm for control. The meta-algorithm has two main components: (1) patient assessment and tuning of control parameters, i.e., algorithmic processes for collection and processing patient data prior to closed-loop operation, and (2) controller warm-up and run-time operation, i.e., algorithmic processes for initializing controller states and managing blood glucose. The simulation-based testing methodology is designed to reveal the conceptual/mathematical operation of both main components, as applied to a large population of in silico patients with type 1 diabetes mellitus.

Impact of sensing and infusion site dependent dynamics on insulin bolus based meal compensation

“…The detection could also benefit from additional information beyond CGM data. Anticipatory approaches, for example, take into account meal prior probabilities based on general or individual eating behavior and sleeping times in order to improve the meal detection (Patek et al, 2009;Cameron et al, 2012). The impact of glucose sensing dynamics on the estimation of meal sizes, and thus allowable bolus sizes, should be analyzed in subsequent work.…”

Development and in silico validation of the PID-algorithm for the artificial pancreas with intraperitoneal insulin delivery

“…Первый доклинический этап тестирования эффективности и безопасности работы управляющего алгоритма принято проводить в условиях программной симуляции. Исследователи из университетов Вирджинии (США) и Падуи (Италия) разработали метаболический симулятор UVA/Padova для облегчения разработки алго-ритмов ИПЖ и их виртуального тестирования (в условиях in silico), которое по одобрению Food and Drug Administration (FDA, США) может заменить этап доклинического те-стирования с участием лабораторных животных [18][19][20].…”

Of rodents and men: Species-specific glucose regulation and type 2 diabetes research

“…this physiologically based model was generated using quantitative information such as plasma glucose concentration, glucose and insulin fluxes previously obtained from normal and T2DM human subjects, and this guided the development of closed-loop glucose control via implantable insulin pumps. Based on the predictive value of the "artificial pancreas control algorithms" computational model, the FDA approved closed-loop insulin pumps as a substitute for animal testing (Dalla Man et al, 2007;Kovatchev et al, 2009;Patek et al, 2009). This is merely one example that "realistic computer simulation is capable of providing invaluable information about the safety and the limitations of closed-loop control algorithms, guiding clinical studies, and out-ruling ineffective control scenarios in a cost-effective manner" (Kovatchev et al, 2009).…”