Glucose Prediction Using Segmented Machine Learning Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies face challenges in accurately predicting physiological parameters, such as blood glucose levels, over a prediction horizon, which is crucial for managing conditions like diabetes to prevent hypoglycemia or hyperglycemia.
Innovation Solution
The use of processor-implemented methods and systems that apply machine learning models to predict blood glucose levels. These models are trained to determine whether the glucose level will cross specific thresholds within a prediction horizon, allowing for the generation of alerts for hypoglycemia or hyperglycemia.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Duration of action of moving object
If a first glucose prediction model is applied for a first prediction horizon to obtain a first predicted glucose value, then the prediction covers a longer time range, but the accuracy for shorter-term predictions deteriorates
Solution Approach 1:
The patent segments the prediction task into two distinct models: a first glucose prediction model for longer prediction horizons and a second glucose prediction model for shorter prediction horizons. Each model is optimized for its specific time range, with the second model receiving the first predicted glucose value as input along with at least one glucose concentration value, enabling accurate short-term predictions while maintaining the capability for long-term forecasting through the first model.
2Reliability
If machine learning models are used to predict blood glucose levels, then the ability to detect future hypoglycemia or hyperglycemia improves, but the complexity of the system increases
Solution Approach 1:
The patent introduces an intermediary processing layer that receives glucose concentration values from sensors and feeds them into machine learning models. The first predicted glucose value serves as an intermediary output that is then combined with at least one glucose concentration value as input to the second glucose prediction model, creating a structured data flow that manages system complexity while maintaining high detection reliability for hypoglycemia andhyperglycemia events.
Data Source
AI summary
Disclosed herein are techniques related to predicting a physiological condition of a user. In some embodiments, the techniques may involve obtaining one or more glucose concentration values measured from a user; applying, to the one or more glucose concentration values measured from the user, a first glucose prediction model for a first prediction horizon; obtaining, based on applying the first glucose prediction model, a first predicted glucose value of the user; and predicting a second predicted glucose value of the user for a second prediction horizon that is less than the first prediction horizon, based on the first predicted glucose value and at least one glucose concentration value of the one or more glucose concentration values. In some scenarios, the physiological condition may include, for example, hypoglycemia or hyperglycemia.