Management of diabetes using adaptive control

Abstract

The review focuses on adaptive systems for insulin treatment in type 1 diabetes. The review consists of two parts. First, adaptive approaches are described, which exploit infrequent glucose measurements (four to seven measurements per day). Second, adaptive approaches are described, which exploit frequent or continuous glucose measurements (every hour or more often). Each part is further divided into two subparts separating off-line and on-line adaptive techniques. The latter represents treatment strategies, which rely on continuous re-assessment of the glucoregulatory system. The former refers to treatment strategies, which are fixed for a day or longer and are revisited from time to time. It is concluded that the role of adaptive approaches will increase as new continuous glucose-sensing monitors reach the market.

Low-complexity MISO models of T1DM glucose metabolism

“…The set-up is that of a basal-bolus therapy, involving impulsive control variables, namely insulin shots and meal carbohydrates, administered several times over the course of the day at irregularly spaced time instants. We remind the reader in passing that this framework differs from most of the proposed strategies to manage diabetes in an automated fashion [4], [5], [21], [22], [23], [24] nevertheless, it is the most widespread approach among the diabetics to cure their disease and in the authors' opinion it represents a viable path worth pursuing and investigating. The parameters in the models are linked to clinical variables.…”

Impulsive Predictive Control of T1DM Glycemia: An <italic>In-Silico</italic> Study

“…The set-up is that of a basal-bolus therapy, involving impulsive control variables, namely insulin injections and meal carbohydrates, administered several times over the course of the day at irregularly spaced time instants. We remind the reader in passing that this framework differs from most of the proposed strategies to manage diabetes in an automated fashion [9], [12], [13], [21], [22], [23] in which glycemia is regulated only with a continuous insulin infusion pump, neverthless, it is the most widespread approach among the diabetics to control their disease. Previous attempts at producing impulsive control signals include approximation of the continuous insulin signal from a model predictive controller [24], [25].…”

Archives of Control Sciences

“…So far, the problem of glycemia control has been addressed in a respectable number of papers, while exploiting various more or less sophisticated strategies starting from simple PID controllers [11][12][13], adaptive approaches [ [14][15][16], robust control [17], and concluding with the predictive control as perhaps the most promising one [18]. Some of the most significant and closely related works will be briefly examined in this section.…”