The next generation of artificial pancreas control algorithms

Area of Science:

• Biomedical Engineering
• Control Theory
• Endocrinology

Background:

• Wearable artificial pancreas (AP) systems aim to improve diabetes management by automating glucose control.
• Current AP control algorithms, based on simple models, lack efficiency and reliability.
• Diabetic individuals face significant complications and reduced quality of life due to glucose dysregulation.

Purpose of the Study:

• To propose corrections to current research directions for developing effective AP systems.
• To highlight the need for advanced control strategies and improved physiological models.
• To suggest hardware advancements for enhanced AP performance.

Main Methods:

• Recommending model predictive controllers (MPCs) that utilize patient-specific physiological models.
• Advocating for whole-body physiologically based pharmacokinetic-pharmacodynamic (PK-PD) models.
• Identifying necessary improvements in diabetes modeling, specifically integrating hypothalamus-pituitary-adrenal (HPA) axis and gastrointestinal (GI) tract submodels.

Main Results:

• Current simple models are insufficient for reliable AP control.
• Physiologically based PK-PD models offer the best potential for successful AP systems.
• Enhancing HPA axis and GI tract submodels is critical for practical AP models.

Conclusions:

• Advanced control algorithms and accurate physiological models are essential for AP development.
• Integrating detailed HPA axis and GI tract submodels requires collaborative efforts.
• Hardware innovations, such as dual insulin and glucagon delivery, can improve AP system robustness.