CHoKI-based MPC for blood glucose regulation in Artificial Pancreas
Модельно-прогнозирующее управление на основе CHoKI для регулирования уровня глюкозы в крови в искусственной поджелудочной железе
2023-01-01
SCID: 54.1/cnsgnehp
Discuss with AI
CHoKI-based MPCUVA/Padova simulatorartificial pancreasblood glucose regulationinsulin-on-board constraints
Figures from the paper
Abstract (AI)
This work presents a Model Predictive Control (MPC) algorithm for the artificial pancreas able to autonomously manage basal insulin injections in type 1 diabetic patients. The MPC goal is to maintain the blood glucose inside the safe range (70-180 mg/dL) acting on the insulin amount, using a model to make predictions of the system behavior and satisfying operational constraints. The complexity of diabetes complicates the identification of a general physiological model, so a data-driven learning method is proposed, the Componentwise Hölder Kinky Inference (CHoKI), leading to customized controllers. For the data collection phase and also to test the proposed controller, the FDA-accepted UVA/Padova simulator is exploited. The final results are promising since the proposed controller reduces the time in hypoglycemia if compared to the standard constant basal insulin therapy, satisfying also the time in range requirements.
Key Findings
1
A CHoKI-based, data-driven MPC autonomously regulates basal insulin delivery for type 1 diabetes in an artificial pancreas.
2
Including IOB constraints produced a conservative controller that reduced hypoglycemia time compared with control without IOB constraints.
3
Patient-specific CHoKI models provide customized control without requiring a general physiological diabetes model.
4
Testing with UVA/Padova virtual patients included insulin-sensitivity variability and physical-activity sessions.
5
The controller targets the safe blood-glucose range of 70–180 mg/dL while accounting for insulin-on-board and imposed insulin constraints.
Research Object
artificial pancreas for autonomous basal insulin delivery in type 1 diabetic patients, evaluated using virtual patients from the UVA/Padova simulator
Research Subject
MPC-based blood glucose regulation, including maintenance within 70–180 mg/dL while accounting for insulin-on-board constraints, insulin-sensitivity variability, and physical activity
Publication Details
Publication Date
2023-01-01
Journal
Publisher
ISSN
Cited by
12
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai4
Safety Constraints in an Artificial Pancreatic β Cell: An Implementation of Model Predictive Control with Insulin on Board2009
Circadian Variability of Insulin Sensitivity: Physiological Input for In Silico Artificial Pancreas2014
Artificial Pancreas: Model Predictive Control Design from Clinical Experience2013
Adaptive Zone Model Predictive Control of Artificial Pancreas Based on Glucose- and Velocity-Dependent Control Penalties2018