Generative Metabolic Prediction Models Using Continuous Glucose Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current healthcare systems face challenges in providing continuous data for metabolic values, leading to limited treatment options and recommendations based on sporadic readings, particularly in managing chronic conditions like diabetes, which often rely on isolated glucose measurements.
Innovation Solution
A computer-implemented method using a generative machine learning model trained with continuous glucose monitoring data to predict metabolic values, incorporating glycemia risk index and time in range metrics, and historical data to generate actionable insights and recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If sporadic glucose readings are used for diabetes treatment, then device complexity is reduced, but measurement precision and reliability of treatment decisions deteriorate
Solution Approach 1:
The patent implements continuous glucose monitoring that collects glucose level data continuously over time rather than relying on sporadic readings. This continuous data collection enables the machine learning model to learn patterns and generate accurate predictions, resolving the contradiction by maintaining continuous monitoring action without requiring complex intervention at each measurement point.
Solution Approach 2:
The system performs preliminary data collection and model training in advance by accumulating historical glucose data and training the machine learning model beforehand. This preliminary action enables accurate predictions to be generated when needed, improving measurement precision without requiring complex real-time processing during critical decision moments.
2Measurement precision
If continuous glucose monitoring data is collected and processed, then prediction accuracy improves, but loss of time for data processing and model training increases
Solution Approach 1:
The patent performs model training in advance using historical data before predictions are needed. By completing the computationally intensive training process beforehand, the system minimizes data processing time during actual prediction scenarios while maintaining high accuracy through the trained model's ability to quickly generate predictions from continuous monitoring data.
3Reliability
If more historical metabolic data is used for training, then model reliability improves, but quantity of data to be processed increases
Solution Approach 1:
The patent extracts and utilizes specifically relevant features from the historical metabolic data such as glucose levels, insulin doses, carbohydrate intake, and exercise information. By selecting and extracting only the most pertinent features rather than processing all raw data, the system achieves high model reliability with reduced data processing requirements.
Data Source
AI summary
A method may receive historical metabolic values for an individual having a first medical condition. A method may provide a first subset of the historical metabolic values to a machine learning model to train a generative machine learning model. A method may generate a first predicted metabolic value based on the first subset of historical metabolic values. A method may calculate a root mean square error (RMSE) between the first predicted metabolic value and a corresponding actual metabolic value of a second subset of historical metabolic values. A method may train the generative machine learning model to minimize the RMSE. A method may generate a trained generative machine learning model based on the training.