Personalized Glucose Prediction Using Bayesian Physiological Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing blood glucose prediction methods for type 1 diabetes are inadequate due to high inter-/intra-patient variability and unknown disturbances, with physiological models being too rigid or complex, and black-box models lacking physiological relevance.
Innovation Solution
A personalized nonlinear physiological model using a Markov Chain Monte Carlo (MCMC) Bayesian estimator to estimate model parameters from patient data, combined with a particle filter for real-time prediction, incorporating subcutaneous insulin absorption, oral glucose absorption, and glucose-insulin kinetics, and a residual error model to account for measurement errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a personalized nonlinear physiological model with multiple parameters is used, then prediction accuracy is improved, but model complexity and difficulty of parameter estimation increase
Solution Approach 1:
The patent applies preliminary action by using a priori population data to pre-estimate certain model parameters before personalization. This allows the system to start with population-level estimates and then refine only the necessary patient-specific parameters, reducing the overall complexity of the parameter estimation process while maintaining high prediction accuracy.
Solution Approach 2:
The patent implements partial action by selectively personalizing only a subset of model parameters rather than all parameters. The system estimates some parameters from population data and personalizes only the most critical patient-specific parameters, thereby achieving good prediction accuracy without the full complexity of complete parameter personalization.
2Reliability
If a Bayesian parameter estimation technique is applied to personalize model parameters, then individual patient dynamics are captured, but computational time and processing requirements increase
Solution Approach 1:
The patent uses preliminary action by pre-computing population-level parameter distributions and storing them as a priori information. This allows the Bayesian estimation process to start with pre-prepared population data, significantly reducing the computational burden during real-time patient-specific parameter estimation while still capturing individual dynamics.
Solution Approach 2:
The patent applies copying by using population parameter distributions as templates that are then adapted to individual patients. Instead of performing full Bayesian estimation from scratch for each patient, the system copies and adapts population-level parameter distributions, reducing computational time while maintaining reliability in capturing individual patient dynamics.
3Measurement precision
If more model parameters are personalized from patient data, then prediction accuracy improves, but data requirements and measurement complexity increase
Solution Approach 1:
The patent implements partial action by personalizing only the most critical model parameters from patient data while obtaining other parameters from population studies. This selective approach achieves good prediction accuracy without requiring extensive patient-specific data collection for all parameters, thereby reducing measurement complexity and data requirements.
Solution Approach 2:
The patent applies universality by using population-level parameter distributions that can serve multiple patients. These universal population parameters reduce the need for extensive patient-specific data collection, as they provide reasonable estimates for parameters that are difficult or expensive to measure individually, while still allowing personalization when patient data is available.
Data Source
AI summary
A method of predicting future blood glucose concentrations of an individual patient includes: selecting an individualized nonlinear physiological model of glucose-insulin dynamics, the selected model having a plurality of model parameters whose values are to be determined; estimating values for each of the model parameters in the plurality of model parameters, a first subset of the model parameters having values estimated from a priori population data and a second subset of the model parameters having values personalized for the individual patient by applying a parameter estimation technique to a priori information and data for the individual patient to obtain a posteriori information; and; applying a nonlinear prediction technique to the selected model using the estimated values for each of the model parameters to obtain a predicted blood glucose concentration of the individual patient at a future time.