Glucose Flux Estimation via Sparse Dictionary Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating glucose rate of appearance from the intestine, endogenous glucose production, and insulin-dependent glucose utilization during meals are invasive, expensive, and prone to errors, lacking a minimally invasive technique to simultaneously measure these glucose fluxes using continuous glucose sensors and insulin infusion measurements.
Innovation Solution
A system employing continuous glucose sensors, dynamic glucose/insulin models, insulin delivery information, meal carbohydrate data, and sparse encoding dictionaries to estimate glucose rate of appearance, endogenous glucose production, and insulin-dependent glucose utilization through convex optimization, eliminating the need for radioactive tracers and intravenous measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple injected radioactive glucose tracers are used to estimate glucose fluxes, then measurement precision is improved, but device complexity and cost increase significantly
Solution Approach 1:
The patent extracts and eliminates the complex radioactive tracer infusion system while retaining the essential measurement function. By using only continuous glucose sensor measurements combined with physiological models, the system removes the need for multiple tracer injections, intravenous access, and complex tracer delivery infrastructure, thereby reducing device complexity while maintaining glucose flux estimation capability
Solution Approach 2:
The patent creates a virtual model of glucose metabolism that replicates the function of physical tracer measurements. By using physiological models (such as minimal models or compartmental models) fitted to routine glucose sensor data, the system generates estimated glucose fluxes that copy the information obtained from radioactive tracers without requiring the actual tracers, thus simplifying the measurement system
2Measurement precision
If multiple injected radioactive glucose tracers are used to estimate glucose fluxes, then measurement precision is improved, but cost increases significantly
Solution Approach 1:
The patent replaces expensive radioactive tracers and complex infusion equipment with inexpensive continuous glucose sensors that can be worn by patients. The sensors provide continuous data over extended periods at low cost, eliminating the need for repeated expensive tracer administrations and complex medical infrastructure, thereby dramatically reducing the overall measurement system cost
3Measurement precision
If multiple injected radioactive glucose tracers are used to estimate glucose fluxes, then measurement precision is improved, but invasiveness increases
Solution Approach 1:
The patent removes the invasive elements of the traditional measurement system, specifically the intravenous catheters and radioactive tracer injections. By relying solely on subcutaneous continuous glucose sensors and mathematical modeling, the system eliminates all invasive procedures while maintaining the ability to estimate glucose fluxes, thereby improving patient comfort and safety
4Measurement precision
If complex protocols with multiple tracers and models are used, then measurement precision is improved, but ease of operation decreases
Solution Approach 1:
The patent enables the measurement system to automatically process routine glucose sensor data through physiological models without requiring manual tracer administration or complex protocol execution by healthcare providers. The system self-calibrates and continuously estimates glucose fluxes using readily available glucose measurements, eliminating the need for specialized training and complex operational procedures
Data Source
AI summary
Method and system for determining glucose flux profiles in plasma during meals using continuous glucose sensors and insulin delivery. A database of plausible glucose flux profiles is encoded in dictionaries using sparse dictionary learning. A constrained Lasso minimization problem is formed that integrates a transport model for a patient with the dictionaries for estimating the glucose fluxes. Meal carbohydrates consumed by a patient is incorporated in the minimization problem through convex constraints. The estimated glucose fluxes resulting from solving the constrained Lasso minimization problem are glucose rate of appearance from the intestine, endogenous glucose production from the liver and insulin dependent glucose utilization. A method for determining patient carbohydrate to insulin ratio at the time of the meal by calculating the area under the curve of the estimated insulin dependent glucose utilization.