Closed-Loop Glucose Control with Multi-Model Predictive Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing closed loop artificial pancreas systems lack the ability to effectively handle abrupt and slow variations in glucose dynamics, often relying on limited control algorithms that do not account for varying insulin pharmacokinetics and patient-specific changes in glucose metabolism.
Innovation Solution
A closed loop system utilizing a multi-model predictive controller (MMPC) that adapts to changes in glucose dynamics by employing multiple state vectors and models, incorporating insulin pharmacokinetic profiles, and adjusting medication delivery parameters in response to glycemic patterns and insulin type, with features like meal boluses and basal rate adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a single fixed model or slowly adapting model is used in control algorithms, then the system is simpler to implement, but it cannot adequately control glucose concentrations when glucose dynamics vary abruptly or change rapidly
Solution Approach 1:
The patent implements a dynamic multi-model architecture where multiple glucose dynamics models are maintained, each representing different physiological states. The system dynamically switches between models based on current glucose trends and physiological conditions, allowing the controller to adapt to abrupt changes in glucose dynamics without requiring complex real-time parameter estimation for a single model.
Solution Approach 2:
The system changes model parameters (such as insulin sensitivity, glucose production rates, and absorption kinetics) based on detected glycemic patterns and insulin pharmacokinetic profiles. This allows the controller to adjust its behavior to match the current physiological state, improving adaptability while maintaining manageable complexity through predefined parameter sets.
2Reliability
If conventional control algorithms are used that do not include insulin pharmacokinetic models, then the control system is simpler, but it cannot effectively handle variations in insulin action and patient-specific glucose metabolism
Solution Approach 1:
The patent introduces an intermediary layer that bridges the control algorithm and the physical insulin-glucose system. This intermediary consists of pharmacokinetic models that translate insulin delivery commands into predicted glucose responses, accounting for absorption delays, distribution patterns, and elimination rates. This layer improves reliability by making the control system aware of insulin dynamics without requiring the entire physiological system to be directly controlled.
Solution Approach 2:
The system creates simplified mathematical copies (models) of the complex insulin-pharmacokinetic-glucose metabolism system. These models replicate the essential dynamics of insulin action and glucose response, allowing the controller to predict and compensate for physiological delays and variations without directly measuring or controlling every physiological parameter.
3Ease of operation
If a closed loop artificial pancreas system is implemented, then patient vigilance requirements are reduced and quality of life improves, but the system must handle both abrupt and slow variations in glucose dynamics which existing algorithms cannot do
Solution Approach 1:
The patent segments the glucose dynamics into multiple distinct models representing different physiological states (fasting, postprandial, exercise, sleep). Each model captures the specific characteristics of that state, allowing the system to handle both abrupt transitions between states and slow variations within states. This segmentation enables the automated system to adapt to diverse conditions without requiring patient intervention.
Data Source
AI summary
The present disclosure relates to systems and methods for controlling physiological glucose concentrations in a patient.