Adaptive Insulin Pump Using Recursive Multivariable Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current insulin pumps are open-loop systems that require manual patient input for meal and activity information, making it difficult to achieve tight blood glucose control due to delays in insulin absorption and variability in glucose-insulin dynamics, and existing closed-loop systems are not commercially available or easily tunable for individual patients.
Innovation Solution
A closed-loop insulin pump system that uses continuous glucose monitoring data and physiological information from an armband monitoring system to develop patient-specific recursive models and adaptive control algorithms, eliminating the need for manual input and providing early hypoglycemia warnings up to 30 minutes in advance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual patient input is required for meal and activity information, then the system can provide personalized insulin control, but the ease of operation deteriorates and patient burden increases
Solution Approach 1:
The system automatically collects physiological data (glucose levels, activity, meal information) and performs insulin dosing calculations without requiring manual patient input. The pump self-adjusts insulin delivery based on real-time sensor data and recursive modeling, eliminating the need for patients to manually enter meal or activity information while maintaining personalized control
Solution Approach 2:
The system continuously monitors glucose levels and other physiological parameters, feeds this information back through recursive multivariable models, and automatically adjusts insulin delivery. This closed-loop feedback mechanism replaces manual patient decisions with automated control based on real-time system state
2Measurement precision
If physiological models are used for predictive control, then blood glucose control accuracy improves, but the device complexity increases due to model tuning requirements
Solution Approach 1:
The system uses recursive least squares estimation to continuously update model parameters based on incoming physiological data. Instead of requiring fixed, pre-tuned parameters, the model parameters adapt automatically to individual patient characteristics and changing conditions, maintaining prediction accuracy without complex manual tuning
Solution Approach 2:
The system transitions from static physiological models to dynamic recursive models that continuously adapt to changing patient states. The model structure remains relatively simple, but the parameters evolve over time based on real-time data, capturing individual variability and metabolic changes without requiring complex model structures
3Ease of operation
If closed-loop control is implemented to eliminate manual input, then ease of operation improves, but the system becomes difficult to tune for individual patients
Solution Approach 1:
The system automatically adapts to individual patient characteristics through recursive parameter estimation. The controller self-tunes by continuously analyzing the relationship between insulin delivery and resulting glucose changes, eliminating the need for manual customization while maintaining individualized control
Solution Approach 2:
The system uses real-time feedback from glucose sensors and physiological monitors to continuously refine control parameters specific to each patient. This adaptive feedback mechanism enables automatic personalization without requiring manual tuning procedures
Data Source
AI summary
A method and device for monitoring or treating patient glucose levels. The device includes a glucose sensor for measuring a glucose level of a patient, a physiological status monitoring system for measuring at least one physical or metabolic variable of the patient, and an automatic controller in communication with the glucose sensor and the physiological status monitoring system. The controller includes a prediction module for automatically predicting a future glucose level using data measured by the glucose sensor and the physiological sensor.