Artificial Pancreas State Estimation for Low-Load Insulin Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing glucose control systems for diabetes management suffer from computational demands, inaccuracies in predictive models, delays in insulin delivery, and reliance on approximate human physiology models, leading to suboptimal insulin dosage and increased risk of hypoglycemia.
Innovation Solution
An artificial pancreas control system using an adaptive filter scheme and Linear Quadratic (LQ) methodology to estimate physiological states from real-time glucose measurements, optimizing insulin dosing recommendations through a dynamic model and Kalman filter to minimize quadratic cost functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Model Predictive Control (MPC) is used to provide glucose control based on prediction, then glucose control accuracy is improved, but computational demand increases significantly
Solution Approach 1:
The patent transforms the complex MPC problem into a simpler Linear Quadratic (LQ) control problem by changing the mathematical parameters and structure. Instead of using full predictive models with multiple constraints and objectives, the invention uses a simplified cost function with quadratic terms that can be solved analytically, reducing computational complexity while maintaining control accuracy.
Solution Approach 2:
The patent extracts only the essential elements needed for glucose control from the full MPC framework. By removing unnecessary predictive components and focusing on the core control objective (minimizing glucose deviation from target), the system achieves adequate performance with significantly reduced computational burden.
2Speed
If control updates are performed at high frequency to improve responsiveness, then glucose control responsiveness is improved, but computational capacity requirements increase
Solution Approach 1:
The patent changes the control formulation from a complex constrained optimization problem to an unconstrained LQ problem with a simple quadratic cost function. This parameter transformation enables analytical solutions that can be computed rapidly at high update frequencies without requiring substantial computational resources.
3Device complexity
If inaccurate predictive models are used to simplify control computation, then device complexity is reduced, but glucose control accuracy deteriorates
Solution Approach 1:
The patent uses a simplified linear quadratic cost function with carefully chosen parameters that capture the essential glucose-insulin dynamics. By transforming the control objective into a mathematical form with specific parameter structures (quadratic terms, weighting factors), the system achieves accurate control results without requiring complex predictive models.
Data Source
AI summary
The invention relates to a methods and systems for determining an insulin dosing recommendation. The invention employs Linear Quadratic methodology to determine the insulin dosing recommendation based on a patient's present physiological state, which is estimated by an adaptive filter methodology employing a dynamic model, which utilizes real time measurements of blood glucose concentration.