According to the International Diabetes Federation, the number of people with diabetes will rise from 284.6 million in 2010 to 438.4 million in 2030. Approximately 10% of diabetic people have type 1 diabetes, characterized by an autoinmune destruction of beta cells in the pancreas, requiring the administration of exogenous insulin for survival. Current insulin therapies, by means of multiple insulin injections or insulin pumps, are not efficient enough to get a good glycemic control.
Technological advances such as insulin pumps and continuous glucose monitoring have been a springboard for research on a wearable “artificial pancreas”, as it has been coined the automatic control of insulin delivery. This system combines a continuous glucose monitor, an insulin pump and a control algorithm “closing the loop”. Significant advances have been done in the field pushed by the Juvenile Diabetes Research Foundation with the creation of the Artificial Pancreas Consortium in 2006 and the Food and Drug Administration with the inclusion of the artificial pancreas as a priority in its Critical Path Innitiative. Several clinical trials have demonstrated the efficacy of such a system in nocturnal glycemic control, reducing hypoglycemic events, in in-clinic settings. Current trials are evaluating the system in situations closer to the patient’s daily life, and soon the first ambulatory trials will be conducted.
However, there exists many challenges to the artificial pancreas. Current technology for continuous glucose monitoring is not accurate enough, specially in hypoglycemia. This is due to the fact that plasma glucose needs to be estimated from measurements of interstitial glucose (calibration algorithm), which have a complex relationship specially under dynamic conditions. A recent study conducted at Universidad Politécnica de Valencia and Universitat de Girona, shows that consideration of dynamic relationships between plasma and interstitial glucose significantly improve accuracy and hypoglycemia detection, compared to current algorithms based mainly on linear regression. More trials are being conducted for algorithms refinement.
Although current control algorithms have shown efficient for nocturnal control, this is not the case for controlling glucose after a meal (postprandial control), either inducing late hypoglycemia due to overcorrection and insulin stacking, or showing poor performace compared to insulin pump therapy due to a too conservative controller tuning. This is a hot topic of research in our group, with the study of postprandial models individualization and optimal experiment design for parametric identification, study of meal absorption and estimation of glucose rate of appearance, development of tools for glucose prediction under intra-patient variability and design of robust and safe control algorithms for postprandial control, among others.
The artificial pancreas is an appealing field for the control engineer and the clinician and many things remain to be done. Better models are needed to describe meal absorption and the impact of nutritional composition, counterregulatory response such as glucagon, effect of exercise and stress on glycemia, circadian variations, allowing better in silico controller validations, as well as better prediction models for control algorithms development. More robust and safe algorithms are needed to reduce insulin stacking and late hypoglycemia. Fault detection and supervision of continuous glucose monitors and insulin pumps is also a promising field, just to name a few. Every small advance in the field will be a great advance in the quality of life of people with type 1 diabetes.
For more information please contact Jorge Bondia: email@example.com
Instituto Universitario de Automática e Informática Industrial http://www.ai2.upv.es/es/index.php
Universitat Politècnica de València