Anticipating the next meal using meal behavioral profiles: a hybrid model-based stochastic predictive control...

Area of Science:

• Biomedical Engineering
• Control Systems Engineering
• Endocrinology

Background:

• Type 1 Diabetes Mellitus (T1DM) management relies on balancing glucose levels through insulin therapy.
• Subcutaneous (SC) glucose sensing and insulin infusion present challenges due to inherent delays, impacting control system stability and responsiveness.
• Model Predictive Control (MPC) offers a way to anticipate and manage glucose fluctuations, but its effectiveness is limited by uncertainty in meal timing.

Purpose of the Study:

• To develop and evaluate a novel control law for T1DM management that accounts for uncertain meal timing.
• To mitigate the destabilizing effects of sensing and actuation delays in SC-SC artificial pancreas systems.
• To improve the safety and aggressiveness of glucose control by incorporating probabilistic meal behavior.

Main Methods:

• Development of a control law utilizing a probabilistic description of patient eating behavior (random meal profile).
• Preclinical in silico trials employing the Dalla Man et al. oral glucose meal model.
• Simulation of the SC-SC control strategy under conditions of uncertain meal arrival.

Main Results:

• The proposed control strategy effectively anticipates meals based on probabilistic behavioral profiles.
• The system demonstrated safe and effective glucose control, even when anticipated meals were skipped.
• The control law successfully accounted for uncertain prior knowledge of meals without compromising patient safety.

Conclusions:

• A novel control law using probabilistic meal profiles can significantly improve artificial pancreas performance in T1DM.
• This approach addresses the critical challenge of mealtime uncertainty in SC-SC diabetes control systems.
• The strategy offers a safe and convenient method for managing glucose levels in Type 1 Diabetes Mellitus.