Adaptive Insulin Dosing Control from Prior Glucose Periods
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current insulin therapies for diabetes require heuristic estimation of correction factors and insulin-to-carbohydrate ratios, which are subject to individual variability and transient changes, leading to inconsistent blood glucose control.
Innovation Solution
An automated system that calculates insulin doses and counter-regulatory agent delivery based on past online operation data, adapting control parameters without user input, using online and offline modes to regulate glucose levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If heuristic estimation methods are used for correction factors and insulin-to-carbohydrate ratios, then the system can operate with minimal user input, but the glucose control consistency deteriorates due to individual variability and transient changes
Solution Approach 1:
The system performs self-learning by automatically analyzing past glucose measurements and insulin delivery data to generate personalized control parameters. The controller autonomously adjusts correction factors and insulin-to-carbohydrate ratios based on observed patterns, eliminating the need for manual user input while improving glucose control consistency through adaptive parameter optimization.
Solution Approach 2:
The system implements feedback mechanisms by continuously monitoring glucose measurements and comparing them with predicted glucose levels. This feedback loop allows the controller to learn from past performance and refine control parameters over time, transforming heuristic estimates into data-driven, personalized parameters that adapt to individual variability and transient changes.
2Adaptability or versatility
If trial-and-error method is used to determine dosing requirements, then the system can be customized to individual needs, but the time required for parameter optimization increases significantly
Solution Approach 1:
The system performs preliminary learning during an initial period by automatically analyzing past glucose measurements and insulin delivery data to establish baseline control parameters. This preliminary action eliminates the need for extended trial-and-error optimization, as the system pre-calculates personalized parameters based on observed patterns before full operation begins.
Solution Approach 2:
The system replaces the manual trial-and-error process with an automated computational model that uses historical data to predict optimal control parameters. This substitution transforms the time-consuming iterative adjustment process into a rapid data-driven calculation, maintaining individual customization while significantly reducing optimization time.
3Adaptability or versatility
If correction factors are manually adjusted over time, then the system can adapt to changing conditions, but the frequency and complexity of user adjustments increases
Solution Approach 1:
The system performs self-adjustment by automatically modifying correction factors and control parameters in response to changing glucose patterns and physiological conditions. The controller monitors glucose measurements and insulin responses to detect changes in metabolic state, then autonomously updates parameters to maintain optimal glucose control without requiring user intervention or complex adjustment procedures.
Data Source
Figure 1~3
Figure 2
Figure 4
AI summary
Apparatus and methods calculate and deliver doses of insulin and optionally glucagon into a subject. Online operation controls delivery of correction doses of insulin automatically in response to regular glucose measurements from a sensor, and offline operation calculates and delivers correction doses based on isolated glucose measurements and information gathered autonomously during preceding online operation. In another aspect, offline operation includes automatically calculating and administering meal doses based on information gathered autonomously during preceding periods of online operation. Both methods include generating relevant control parameters tailored to the individual and continually converged upon and potentially modulated during online operation. The control parameters are employed in real time during periods of offline operation to regulate glucose level without the need for user-provided control parameters such as correction factors and insulin-to-carbohydrate ratios.