Continuous glucose monitoring sensors are crucial for managing blood glucose levels in artificial pancreases, but real-time measurement of insulin and glucagon is currently limited, which affects decision-making.
The paper proposes a nonlinear high-gain observer designed for a bi-hormonal artificial pancreas to estimate unmeasurable states, using a modified intraperitoneal animal model, accommodating for measurement noise and uncertainties.
The developed observer is tested in scenarios with hormone infusions and shows promising results, indicating its potential for enhancing the performance of closed-loop systems in managing Type 1 diabetes by better predicting the body's response to hormone treatments.