Bolus Calculator Using Glucose Convolution for Insulin Dosing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing diabetes management systems lack an efficient and accurate method for calculating insulin bolus recommendations that account for dynamic blood glucose levels and individual patient factors, leading to suboptimal therapy effectiveness.
Innovation Solution
A bolus calculator that convolutes time series of blood glucose values with known insulin pulse responses, considering patient specifics and different types of insulin, to determine the most appropriate insulin bolus, which can be delivered via a drug delivery device.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional diabetes management systems are used, then basic blood glucose monitoring is available, but accurate bolus calculation considering dynamic blood glucose levels and individual patient factors is not achieved
Solution Approach 1:
The system performs preliminary actions by pre-characterizing insulin pulse responses for different insulin types and storing them in a database. It also pre-processes blood glucose time series data. This preparation enables accurate bolus calculation without requiring complex real-time computations during the actual insulin dosing decision.
Solution Approach 2:
The patent introduces an intermediary convolution operation that combines blood glucose time series data with pre-characterized insulin pulse responses. This mathematical operation serves as a mediator between the raw input data (blood glucose measurements) and the desired output (accurate bolus calculation), enabling precise results without direct complex modeling of the entire insulin-glucose system.
2Reliability
If comprehensive patient factors and dynamic blood glucose levels are considered, then therapy effectiveness is improved, but calculation time and processing requirements increase
Solution Approach 1:
The system performs preliminary characterization of insulin pulse responses for different insulin types and stores them in a database. It also pre-processes and stores blood glucose time series data. This preparation enables accurate bolus calculation without requiring complex real-time computations during the actual insulin dosing decision.
Solution Approach 2:
The patent creates a simplified representation (copy) of the complex insulin-glucose system dynamics through pre-characterized pulse response functions. These copied responses capture the essential behavior patterns of different insulins under various conditions, allowing the system to query and combine these pre-computed patterns rather than performing complex simulations in real-time.
3Adaptability or versatility
If multiple types of insulin with different active profiles are supported, then adaptability to patient needs is improved, but device complexity increases
Solution Approach 1:
The patent segments the insulin characterization by creating separate pulse response profiles for different insulin types (rapid-acting, intermediate-acting, long-acting). Each insulin type has its own dedicated response function stored in the database. This segmentation allows the system to handle multiple insulin types by simply selecting the appropriate pre-characterized profile, avoiding the need for a single complex universal model.
Solution Approach 2:
The patent creates a universal framework that can handle multiple insulin types through a single convolution operation. The same mathematical approach and data processing pipeline work for all insulin types; the only variation is which pre-characterized pulse response function is selected from the database. This multi-functional design achieves versatility without proportionally increasing system complexity.
Data Source
Figure 1A~1B
Figure 2
Figure 3
AI summary
The present disclosure relates to a bolus calculator (4.1) for determining a bolus (B) of insulin, the bolus calculator (4.1) having an input configured to be fed with a time series (h(t-τ)) of blood glucose values and to store at least one known pulse response (x(τ)) representing an active profile of at least one insulin, wherein the bolus calculator (4.1) is configured to convolute the time series (h(t-τ)) of blood glucose values with the known pulse response (x(τ)) to obtain the bolus (B)and to a method for calculating a bolus (B) of insulin, comprising feeding a time series (h(t-τ)) of blood glucose values into an input of a bolus calculator (4.1) and storing at least one known pulse response (x(τ)) representing an active profile of at least one insulin in the bolus calculator (4.1), convoluting the time series (h(t-τ)) of blood glucose values with the known pulse response (x(τ)) to obtain the bolus (B).