Towards Safe and Robust Closed-Loop Artificial Pancreas Using Adaptive Weighted PID Control Strategy

Abstract

Artificial pancreas enhances the life experience for diabetic patients by allowing them to live normally with their glucose levels controlled automatically with minimal or no intervention. For closed-loop glucose controllers to be approved for clinical practice, they have to prove safety under all potential scenarios. One of the biggest challenges of closed-loop glucose control is to handle the distortion caused by meal intake. In this paper, we construct an improvement on the Proportional- Integral- Derivative (PID) controller by proposing an adaptive weighted PID strategy for robust glucose control under varying meal conditions. The proposed approach mimics the recommended clinical practice by varying the weights based on the current and previous status. Statistical model checking was conducted to analyze the performance figures and safety properties as compared with the standard PID strategy. The results have shown that the proposed approach outperforms the standard PID control in avoiding postprandial hypoglycemia without sacrificing the glycemic control.