Autoregressive PID Controller for Artificial Pancreas Insulin Delivery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current artificial pancreas systems using PID controllers face challenges in dynamically compensating for insulin delivery delays and inaccuracies, leading to glucose level overshooting, especially during rapid glucose rises from meals.
Innovation Solution
The integration of an autoregressive model with the PID controller to account for insulin absorption lag, allowing for more reliable insulin delivery calculations by averaging results from both models until they converge, thereby optimizing insulin delivery to maintain stable glucose levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If a PID controller is used to regulate blood glucose levels, then the system can automatically control insulin delivery, but glucose level overshooting occurs due to lack of dynamic compensation for insulin delivery delays
Solution Approach 1:
The autoregressive model predicts future glucose levels by analyzing historical glucose data patterns, allowing the system to anticipate glucose excursions before they occur. This predictive capability enables the controller to adjust insulin delivery in advance, preventing overshooting rather than reacting after the fact.
Solution Approach 2:
The system continuously monitors glucose levels and uses this feedback to update both the autoregressive model predictions and PID controller adjustments. The iterative refinement of predictions based on actual measured values creates a closed-loop system that adapts to changing physiological conditions and corrects deviations from target glucose levels.
2Speed
If aggressive PID control is applied to regulate blood glucose, then faster glucose control is achieved, but dynamic compensation for insulin absorption lag is insufficient causing oscillations
Solution Approach 1:
By predicting future glucose levels using the autoregressive model, the system can proactively adjust insulin delivery to account for the known lag in insulin absorption and action. This allows faster response to anticipated glucose changes without causing oscillations, as the controller is prepared in advance rather than reacting to already-occurred deviations.
Solution Approach 2:
The system dynamically adapts control parameters based on real-time glucose measurements and predicted trends. The autoregressive model continuously updates its predictions based on recent glucose patterns, allowing the control strategy to adjust its aggressiveness dynamically - being more proactive when glucose trends indicate upcoming excursions and more conservative when levels are stable.
3Adaptability or versatility
If rapid-acting insulin is used in insulin pumps, then insulin profiles can be changed quickly, but the system is still limited by the speed of insulin absorption and action
Solution Approach 1:
The autoregressive model predicts future glucose levels and recommends insulin delivery adjustments in advance, allowing the system to account for insulin absorption delays by planning delivery timing proactively. This predictive approach effectively compensates for the time lag inherent in rapid-acting insulin without requiring faster insulin pharmacology.
Solution Approach 2:
The autoregressive prediction model acts as an intermediary between glucose sensor measurements and insulin pump delivery commands. It processes the time-delayed insulin effects and translates them into predictive recommendations that account for absorption kinetics, bridging the gap between current insulin delivery and future glucose responses.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
Closed loop control method to control artificial pancreas and artificial pancreas using the same, the method comprising: constructing an autoregressive model by initiatively introducing insulin absorption lag factor, calculating an amount of insulin to be delivered at the current time using the autoregressive model and a PID controller respectively, and tuning parameters of the autoregressive model and the PID controller respectively using the average of the calculation results in order to provide a more accurate prediction of glucose trends and a more desirable amount of insulin delivery.