In Silico Diabetes Simulation Environment for Control Algorithm Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current diabetes management systems, particularly for type 1 diabetes, face challenges in accurately simulating individual glucose-insulin dynamics and sensor errors, limiting the effectiveness of closed-loop control algorithms for outpatient use.
Innovation Solution
A computer simulation environment is developed, incorporating a Glucose Insulin Model (GIM) with individualized parameters for 300 subjects, simulating subcutaneous continuous glucose monitoring and insulin delivery, and accounting for sensor errors to test treatment strategies and control algorithms in a realistic manner.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If population-based diabetes models are used for simulation, then computational efficiency is improved, but individual glucose-insulin dynamics accuracy deteriorates
Solution Approach 1:
The patent creates virtual copies (avatars) of real diabetic patients by transferring their actual physiological parameters, sensor characteristics, and insulin pump settings into simulation models. This allows individualized testing of control algorithms on multiple patient-specific virtual models simultaneously, maintaining both computational efficiency and individual accuracy.
Solution Approach 2:
The simulation system segments the diabetes management system into distinct modular components: glucose dynamics models, sensor error models, insulin pump models, and control algorithms. This modular segmentation enables efficient computational processing while allowing precise individualization of each component's parameters for different patient avatars.
2Device complexity
If simplified sensor and pump models are used, then model complexity is reduced, but realism of simulation deteriorates
Solution Approach 1:
The patent changes the parameters of existing sensor and pump models to match actual device characteristics. Sensor models incorporate specific error patterns, delays, and measurement ranges observed in real continuous glucose monitors. Insulin pump models include actual delivery rates, absorption dynamics, and operational constraints from real devices, enhancing simulation realism without requiring completely new complex models.
3Measurement precision
If individualized patient parameters are implemented, then simulation accuracy for specific patients is improved, but data requirements and model setup complexity increase
Solution Approach 1:
The patent performs preliminary actions by pre-processing real patient data during avatar creation, organizing physiological parameters, sensor characteristics, and pump settings into structured formats ready for simulation. This preliminary data preparation and parameter extraction work is done once per patient, reducing the complexity of setting up individualized simulations for each new patient avatar.
Data Source
Figure 1
Figure 2
Figure 3A~3B
AI summary
A simulation environment for in silico testing of monitoring methods, open-loop and closed-loop treatment strategies in type 1 diabetes. Some exemplary principal components of the simulation environment comprise, but not limited thereto, the following: 1) a "population" of in silico "subjects" with type 1 diabetes in three age groups; 2) a simulator of CGM sensor errors; 3) a simulator of insulin pumps and discrete insulin delivery; 4) an interface allowing the input of user-specified treatment scenarios; and 5) a set of standardized outcome measures and graphs evaluating the quality of the tested treatment strategies. These components can be used separately or in combination for the preclinical evaluation of open-loop or closed-loop control treatments of diabetes.