Glycemia Regulation in Type 1 Diabetes Using Artificial Pancreas Systems: One Decade Review of Positive Control Approaches
Abstract
An artificial pancreas (AP) is a biomedical device that mimics the natural glucose-regulation process carried out by a healthy pancreas. This device consists of three parts: a continuous glucose monitor (CGM), a control algorithm, and an insulin pump. These three parts work together by continuously monitoring blood glucose concentration (BGC) and adjusting insulin delivery to maintain normoglycemia while minimizing the occurrence of hypoglycemia and hyperglycemia. While several review studies have addressed control strategies for AP systems, limited attention has been given to the fundamental positivity constraint of insulin delivery and its systematic impact on controller design and performance. This paper presents a review of existing studies that primarily focus on the design of different control strategies to address the positivity constraint and its implications for BGC regulation across a variety of commonly used type 1 diabetes mellitus (T1DM) models. In contrast to existing reviews, this study explicitly connects control design methodologies to positivity management and assesses their robustness and safety implications. The positive control concept was derived from the intrinsic unidirectional nature of insulin (control input) delivery, as insulin can only be infused, not withdrawn. A classification of control strategies is proposed based on their effectiveness in managing positivity and input constraints, grouping them into classical, model-based, and intelligent control approaches. In addition, the performance of these control strategies is compared under various T1DM model dynamics and physiological conditions, especially during meal disturbances and open-loop intervals when insulin delivery is suspended. Such limitations highlight the need for more robust, safe, and efficient glycemic regulation control strategies, as well as key directions for the future design of AP systems.