Feasibility of Automating Insulin Delivery for the Treatment of Type 1 Diabetes

Abstract

O ptimal treatment of type 1 diabetes should achieve normoglycemia at all times, without risk of hypoglycemia. Such a treatment should dramatically reduce or prevent diabetes complications and significantly improve patients' quality of life. This goal may be accomplished through pancreatic or islet cell transplantation, but availability of these tissues is limited, survival and function are unpredictable, and longterm immunosuppressive therapy is required (1). The potential for an automated closed-loop system, or artificial ␤-cell, to achieve round-the-clock glycemic control, has not been fully explored.An artificial ␤-cell requires a glucose sensor, an insulindelivery pump, and an algorithm for calculating insulin delivery. Technological and scientific advances have made sensors and pumps available, but linking the two as a "closed loop" has been challenging (2). Lingering questions remain regarding the suitability of different glucosesensing sites (subcutaneous versus intravascular), insulindelivery sites (subcutaneous versus intravascular versus intraperitoneal), and sensor reliability. In addition, no one algorithm has been universally accepted as optimal for insulin delivery (3).Herein, we describe the feasibility of achieving glycemic control in patients with type 1 diabetes using a system comprised of a subcutaneous glucose sensor, an external insulin pump, and an algorithm emulating the ␤-cell's multiphasic glucose-induced insulin release (4 -6). ). Subjects had been treated with continuous subcutaneous insulin infusion (CSII) using Lispro insulin (Lilly, Indianapolis, IN) for at least 6 months before study enrollment and were required to have an HbA 1c Ͻ9%. Data from a previously published study (7) characterizing insulin secretion over a 24-h period in nondiabetic subjects are included for comparison of the glucose profiles (n ϭ 17) obtained with a similar diet. The study was approved by the University of California, Los Angeles Institutional Review Board, and all patients gave written informed consent. RESEARCH DESIGN AND METHODSGlycemic control under CSII therapy was characterized over a 3-day outpatient period using a continuous glucose monitoring system (CGMS) (Medtronic MiniMed, Northridge, CA). The CGMS records sensor current every 5 min and glucose profiles are obtained retrospectively (8). Patients were instructed to keep their daily routine but to take a minimum of seven fingerstick blood glucose readings per day (preprandial and 2-h postprandial and at bedtime) with their home glucose meters. Patients were also instructed to record meal carbohydrate content, physical activity, and any hypoglycemic episodes or supplemental carbohydrate in a logbook.To evaluate the closed-loop insulin delivery system, patients were admitted to the general clinical research center at ϳ5:00 P.M., and their insulin pump was replaced with a Medtronic 511 Paradigm Pump capable of communicating telemetrically with a laptop computer. Two subcutaneous glucose sensors were inserted in the abdominal area and connected to...

Algorithms for a Closed-Loop Artificial Pancreas: The Case for Proportional-Integral-Derivative Control

“…Peak postprandial glucose levels declined from ~250 to 180 mg/dl, and levels 3-4 h postprandial became more stable. Meal responses obtained with MPC by teams comprising investigators at the University of Virginia and Padova and at Boston University and Massachusetts General Hospital [13][14][15][16] (Figure 1B) generally showed the peak breakfast response to be similar to the initial PID study 2 . Still, comparing meal responses across studies can be confounded by differences in meal size, and averaging data inevitably results in peak postprandial values lower than the average of individual peak values.…”

“…The group at Boston University/Massachusetts General Hospital has advocated the use of glucagon throughout development of their MPC algorithm, 15,16 whereas the PID algorithm developed at Medtronic, [2][3][4] and variations evaluated at Boston Children's Hospital/Joslin Diabetes Center, 5,6 have been developed for use with insulin alone. The use of glucagon is not strictly speaking an MPC versus PID question, as glucagon can be used with PID control.…”

“…This article will focus on the PID algorithms I have developed with my colleagues. [2][3][4][5][6][7][8] This version did not originate as a "PID" algorithm per se, 9 but rather as a model of β-cell insulin secretion, 10,11 referred to as "physiologic insulin delivery." It was, however, first implemented 2 using only the proportional, integral, and derivative terms needed to fit the β-cell response to a hyperglycemic clamp.…”

Insulin delivery and nocturnal glucose control in children and adolescents with type 1 diabetes

“…Two main categories of control algorithms have been employed, the proportional-integral-derivative (PID) controller [48,49], and the model predictive controller (MPC) [50]. Other approaches include controllers based on fuzzy logic (‘MD logic’) [51] or a combination of MPC and PID for insulin and glucagon codelivery [52].…”

A feedforward–feedback glucose control strategy for type 1 diabetes mellitus

“…A variety of PID control strategies have been developed for diabetes and described in survey papers and articles [7][8][9]11,[37][38][39]. Most of the related evaluations have been based on simulation studies but experimental applications to dogs and humans have also been reported.…”