Glucose Estimation Using Discrete Measurements and ML
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Continuous glucose monitoring systems are expensive and not affordable for many patients, making it difficult for them to manage their blood glucose levels effectively, especially for those who cannot wear a continuous glucose monitor regularly due to cost or comfort issues.
Innovation Solution
A processor-implemented method and system that uses discrete blood glucose measurements, activity data, and contextual information to generate estimated blood glucose values through machine learning-based models, allowing for continuous glucose monitoring without the need for a continuous glucose sensor on a regular basis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If continuous glucose monitoring systems are used, then glucose level monitoring accuracy and continuity are improved, but system cost and affordability deteriorate
Solution Approach 1:
The patent creates a virtual copy of continuous glucose monitoring functionality by using machine learning models to generate estimated glucose values from intermittent fingerstick measurements and sensor data. This virtual CGM system replicates the benefits of actual CGM without requiring continuous sensor wear, thereby reducing cost while maintaining monitoring accuracy.
Solution Approach 2:
The system replaces expensive, reusable continuous glucose sensors with a combination of inexpensive fingerstick test strips and computational algorithms. The fingerstick measurements are taken intermittently rather than continuously, reducing the need for expensive durable sensors while providing sufficient glucose monitoring capability at lower cost.
2Measurement precision
If continuous glucose sensor is worn regularly, then continuous glucose monitoring capability is improved, but patient comfort and wearability deteriorate
Solution Approach 1:
The system replaces continuous sensor wear with periodic fingerstick measurements supplemented by machine learning-based glucose estimation. Instead of wearing a sensor continuously, patients perform intermittent fingersticks at scheduled intervals or when glucose levels need checking, significantly improving comfort while maintaining monitoring capability through the virtual CGM algorithm.
Solution Approach 2:
The virtual continuous glucose monitoring system creates a computational model that replicates continuous glucose tracking without requiring physical continuous sensor attachment. The machine learning algorithm processes intermittent measurements to generate continuous glucose profiles, providing the monitoring capability without the discomfort of constant wear.
3Ease of manufacture
If machine learning-based glucose estimation is used, then system cost is reduced, but measurement precision may deteriorate
Solution Approach 1:
The system uses machine learning algorithms as an intermediary that processes multiple input data sources (fingerstick glucose values, sensor measurements, patient activity data, meal information) to generate accurate glucose estimates. This computational mediator synthesizes information from various inexpensive sources to achieve precision that would be difficult to obtain from any single low-cost source alone.
Solution Approach 2:
The glucose estimation system combines multiple data types and measurement sources into a composite virtual glucose profile. By integrating fingerstick measurements, sensor data, contextual information about meals and activity, and machine learning predictions, the system creates a composite view of glucose levels that compensates for the limitations of individual inexpensive measurement methods.
Data Source
AI summary
Disclosed herein are techniques related to glucose estimation without continuous glucose monitoring. In some embodiments, the techniques may involve receiving input data associated with a user. The input data may comprise discrete blood glucose measurement data associated with the user, activity data associated with the user, contextual data associated with the user, or a combination thereof. The techniques may also involve using an estimation model and the input data associated with the user to generate one or more estimated blood glucose values associated with the user.