Adaptive Insulin Delivery Algorithm Tuning via Historical Data Assessment

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current medication delivery algorithms, such as automatic insulin delivery algorithms, face delays in improving glucose control outcomes due to the processing of significant amounts of data, which hinders real-time assessment and adaptation.

Innovation Solution

A controller for a drug delivery system that processes historical blood glucose measurement and insulin delivery data to calculate expected insulin delivery amounts, executes multiple advisory mode algorithms, and updates algorithm settings based on the most suitable output, enabling real-time assessment and adaptive improvements.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If significant amounts of insulin delivery data and blood glucose measurement data are processed in a stepwise process, then the AID algorithm can be assessed and improved, but the time delay for each AID algorithm to improve its behaviors increases significantly

Engineering Contradiction:

Improveassessment accuracyVSAvoidimprovement delay

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system pre-calculates expected insulin delivery amounts and pre-executes multiple advisory mode algorithms with different parameters before actual delivery cycles complete. This allows the assessment process to begin earlier, using historical data that is already prepared and processed, thereby reducing the overall time delay while maintaining comprehensive data analysis for accurate assessment.

Inventive Principle:

Principle #10Preliminary action

2Reliability

If multiple advisory mode algorithms with different parameters are executed and evaluated, then the selection of optimal algorithm settings improves glucose control outcomes, but the computational complexity and processing time increase

Engineering Contradiction:

Improveglucose control outcomeVSAvoidalgorithm evaluation complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The assessment process is divided into separate modular components: data acquisition module, expected delivery calculation module, multiple advisory algorithm execution module, evaluation module, and selection module. Each module handles a specific task independently, allowing for parallel processing of multiple algorithms while maintaining clear separation of concerns. This segmentation reduces overall system complexity and enables more efficient computation.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system executes multiple advisory mode algorithms that differ in their parameter settings (such as insulin sensitivity factors, carbohydrate ratios, and target glucose levels). By systematically varying these parameters across different algorithm instances, the system can evaluate performance across a range of conditions and select the optimal parameter set for the specific patient's needs, improving glucose control while managing complexity through structured parameter exploration.

Inventive Principle:

Principle #35Parameter changes

3Speed

If real-time assessment based on near term retrospective insulin delivery history is implemented, then the responsiveness of AID algorithm improvement increases, but the data processing requirements and computational load increase

Engineering Contradiction:

Improvereal-time assessment speedVSAvoiddata processing energy

Core Design Contradiction:

SpeedVSUse of energy by moving object

Solution Approach 1:

The system extracts and processes only the most relevant near-term retrospective data elements required for assessment, rather than analyzing complete historical datasets. By identifying and extracting key parameters such as recent glucose measurements, corresponding insulin deliveries, and contextual factors from the past delivery history, the system reduces computational load and energy consumption while maintaining real-time assessment capability focused on the most impactful data points.

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS20230381414A1Assessment of past insulin delivery outcomes to automatically tune MDA systems

Publication Date: 2023.11.30 INSULET CORP

AI summary

Disclosed are techniques, devices and systems that obtain a glucose measurement history and a liquid drug delivery history. An expected drug delivery amount may be calculated based on the obtained glucose measurement history and the obtained liquid drug delivery history. A processor may calculate a plurality of respective drug delivery amounts implemented using different advisory mode algorithms. A respective advisory drug delivery amount of the plurality of respective advisory drug delivery amounts may be selected by the processor. A recommendation may be generated based on the selected respective advisory drug delivery amount.