Dynamic Prediction Horizon for Insulin Delivery Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current model-based automatic insulin delivery systems face challenges in accurately modeling and projecting glycemic disturbances due to high uncertainty in glucose dynamics, leading to inefficiencies in attaining insulin delivery targets.
Innovation Solution
An insulin delivery controller is implemented with a processor configuration that processes glucose data to determine glycemic disturbances and their rate of change, generating command signals to dynamically reshape glycemic disturbances within a prediction horizon, thereby adjusting insulin delivery dosages and rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If model-based automatic insulin delivery systems are used to control insulin dosing, then glycemic control can be improved, but high uncertainty in glucose dynamics leads to inaccurate modeling and projection of glycemic disturbances
Solution Approach 1:
The system dynamically changes parameters including the prediction horizon length and glycemic disturbance reshaping based on the glucose rate of change. When glucose is rising rapidly, the prediction horizon is extended and disturbance is increased to account for delayed insulin action. When glucose is stable or falling, the horizon is shortened and disturbance reduced, improving accuracy while adapting to current glycemic conditions
Solution Approach 2:
The system implements dynamic adjustment of the prediction horizon and glycemic disturbance parameters based on real-time glucose rate of change. This transforms the static model into a dynamic system that adapts its complexity and parameters according to the current glycemic state, allowing accurate modeling without fixed complex structures
2Measurement precision
If iterative calculations are performed to attain insulin delivery targets, then glycemic control precision can be improved, but processing time and computational resources are increased
Solution Approach 1:
The system performs preliminary action by proactively adjusting insulin delivery based on predicted glycemic disturbance before hypoglycemia or hyperglycemia occurs. By using the dynamically adjusted prediction horizon and glycemic disturbance reshaping, the system anticipates future glycemic states and takes preventive action, reducing the need for iterative corrections and achieving targets more efficiently
Solution Approach 2:
The system skips unnecessary iterative calculations by using the dynamically optimized prediction horizon. When the glucose rate of change indicates stable conditions, the shortened horizon allows the system to reach the insulin delivery target with fewer calculation iterations, reducing processing time while maintaining precision
3Reliability
If insulin delivery dosage is increased to compensate for glycemic uncertainty, then hyperglycemia can be prevented, but the risk of hypoglycemia increases
Solution Approach 1:
The system applies local quality by adjusting the glycemic disturbance parameter locally based on the specific glucose rate of change condition. Rather than uniformly increasing insulin to prevent hyperglycemia, the system selectively increases disturbance only when glucose is rising, and reduces or eliminates it when glucose is stable or falling, preventing hypoglycemia while maintaining hyperglycemia prevention
Solution Approach 2:
The system applies preliminary anti-action by using glycemic disturbance reshaping to counteract potential hyperglycemia before it occurs, while simultaneously preventing the opposite harmful effect. The dynamic adjustment ensures that anti-action is applied only when needed (when glucose is rising), avoiding the creation of harmful hypoglycemia
4Device complexity
If fixed prediction horizon is used in insulin delivery control, then system simplicity is maintained, but adaptability to varying glucose dynamics is reduced
Solution Approach 1:
The system transitions from a fixed prediction horizon to a dynamic one that adjusts based on glucose rate of change. This allows the system to adapt its time horizon to match the current glycemic dynamics - extending it when glucose is changing rapidly and shortening it when stable - improving adaptability without requiring a completely complex system architecture
Solution Approach 2:
The system changes the prediction horizon parameter dynamically based on glucose rate of change conditions. This simple parameter adjustment mechanism provides adaptability to varying glucose dynamics while maintaining system simplicity, as it only requires monitoring glucose rate and adjusting the horizon length accordingly
Data Source
AI summary
Embodiments can relate to an insulin delivery controller which implements a processor configuration to efficiently attain an insulin delivery target. The insulin delivery controller can include a processor and a memory associated with the processor. The processor can process glucose data received from the memory, including a data representation of glycemic disturbance (d(t)). The processor can determine a glucose rate of change (G′(t)). The processor can generate a command signal to dynamically reshape a glycemic disturbance within a prediction horizon of the insulin delivery controller according to the G′(t). The processor can generate an insulin command signal for an insulin delivery unit to adjust an insulin delivery dosage amount and/or an insulin delivery dosage rate.