Dynamic Insulin Safety Constraints Using Delivery History Divergence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current closed-loop insulin delivery systems face limitations in dynamically adjusting insulin delivery based on individual user physiology, often relying on rigid safety constraints and single data streams, which can lead to inadequate glucose control and increased risks of hypoglycemia or hyperglycemia.
Innovation Solution
Implementing a dynamic, personalized approach with 'soft' safety constraints and advanced glucose prediction models that incorporate real-time data from multiple sources, including continuous glucose monitors, user interactions, and physiological parameters, to adjust insulin delivery algorithms and safety limits in real-time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If hard safety constraints with fixed limits on insulin delivery are applied to all users, then system safety is improved, but optimization of insulin delivery specific to each user is unduly limited
Solution Approach 1:
The patent transforms fixed, static safety constraints into dynamic, adaptive constraints that evolve over time. The system implements a softening mechanism where hard safety limits are gradually relaxed based on accumulated delivery history and demonstrated user safety, allowing the system to adapt from conservative fixed limits to personalized dynamic limits that optimize insulin delivery for each user.
Solution Approach 2:
The system changes the parameter of safety constraints from fixed values to variable values that depend on delivery history and user responses. By modifying the constraint parameters based on accumulated data about user physiology and responses to insulin deliveries, the system achieves both safety and user-specific optimization.
2Device complexity
If rigid models of blood glucose and insulin interactions are used to calculate recommendations, then computational simplicity is improved, but predictive capabilities are unduly limited
Solution Approach 1:
The system incorporates feedback loops where actual user responses to insulin deliveries are measured and used to refine predictive models over time. By continuously comparing predicted versus actual blood glucose responses and adjusting models based on delivery history patterns, the system improves predictive accuracy while maintaining computational tractability through iterative learning.
3Device complexity
If closed-loop insulin delivery systems rely on single data stream from continuous glucose monitor, then system complexity is reduced, but ability to adjust delivery based on user physiology is limited
Solution Approach 1:
The system uses the user's own delivery history and response patterns as additional data sources, eliminating the need for external sensors. By mining information from existing delivery records and glucose responses, the system self-generates physiological insights without requiring additional body-mounted sensors or increased hardware complexity.
Data Source
AI summary
Disclosed are a device, system, methods and computer-readable medium products that provide techniques to implement functionality to receive blood glucose measurements over a period of time. An average of missing blood glucose measurement values may be maintained over a predetermined time period. A count of a number of missing blood glucose measurement values over a period of time may be established. A controller may calculate a divergence of the number of missing blood glucose measurement values over the period of time from the average of missing blood glucose measurements over the predetermined time period. Based on a value of the divergence, a determination that a safety constraint for delivery of insulin is to be reduced. The safety constraint may be reduced by a predetermined percentage. An instruction to deliver an insulin dosage may be generated according to the reduced safety constraint may be forwarded to a wearable drug delivery device.