System and Method for an Artificial Pancreas With Multi-Stage Model Predictive Control

This patent describes an artificial pancreas system that uses a layered control strategy to dose insulin automatically for people with diabetes. It comes from the University of Virginia Patent Foundation, which manages inventions from the University of Virginia.

The Problem

For people with type 1 diabetes, keeping blood sugar in a safe range is a constant balancing act. Too little insulin lets glucose run high, causing long-term harm; too much can drive it dangerously low. An artificial pancreas tries to automate this by pairing a continuous glucose monitor with an insulin pump and a control algorithm that decides dosing. The hard part is that the body responds to insulin slowly and unpredictably, and factors like meals, exercise, and stress all shift the target. A controller that reacts too aggressively risks lows, while one that is too cautious lets sugar drift high. The challenge is a control strategy that anticipates changes and manages risk across different timescales at once.

What This Invention Does

The patent sets out a system and method for an artificial pancreas driven by multi-stage model predictive control, or MPC. Model predictive control uses a mathematical model of how a patient’s glucose responds to insulin to look ahead and choose doses that steer glucose toward target while respecting safety limits. The multi-stage design applies this prediction in more than one layer or over more than one horizon, so the system can plan longer-term glucose trends while still reacting to immediate changes, balancing responsiveness against the risk of going too low. The system takes readings from a glucose sensor, runs the layered predictive calculation, and commands the insulin pump accordingly. The filing covers both the overall system and the control method, aimed at tighter, safer automated glucose control.

Key Features

• Predictive dosing. Model predictive control uses a glucose response model to choose insulin doses that look ahead.
• Multi-stage design. Prediction is applied in more than one layer or horizon to balance long-term and immediate goals.
• Risk management. The layered approach guards against driving glucose dangerously low.
• Closed loop. The system reads a glucose sensor and commands an insulin pump automatically.
• System and method. The filing covers both the device arrangement and the control algorithm.

Who Is Behind It

The applicant is the University of Virginia Patent Foundation, which manages and licenses technology developed at the University of Virginia, a longstanding centre of artificial pancreas research. The named inventors are Marc D. Breton, Jose Garcia-Tirado, Patricio Colmegna, and John Corbett. This filing is a divisional application from an earlier related Australian application.

Why It Matters

Automated insulin delivery has already changed daily life for many people with type 1 diabetes, and better control algorithms are what push these systems toward tighter, safer, more hands-off performance. A predictive strategy that plans across timescales can hold glucose closer to target while lowering the risk of dangerous lows. Protecting the technology in Australia supports its development and licensing for use in the region.

Related Concepts

• Artificial pancreas – the automated insulin delivery system this patent describes.
• Model predictive control – the forward-looking control method at its core.
• Type 1 diabetes – the condition the system is designed to manage.
• Continuous glucose monitoring – the sensing that feeds the controller.
• Insulin pump – the device that delivers the doses the algorithm calculates.

---

AU 2026201994 was published in the Australian Official Journal of Patents on 2 April 2026 and is open for public inspection. Patent applications represent inventions that are sought to be protected and do not necessarily reflect commercially available products.