Retrospective Glucose Sensor Calibration via Wavelet Decomposition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current continuous glucose monitoring systems require frequent external calibration using finger sticks, which are painful, costly, and prone to errors, and lack reliable self-calibration and diagnostics to distinguish between sensor failures and physiological changes.
Innovation Solution
A method employing physical sensor electronics, a microcontroller, and discrete wavelet decomposition to calculate glucose levels using electrode current values, combined with machine learning models and Electrochemical Impedance Spectroscopy for sensor diagnostics and redundancy, reducing the need for external calibration and improving reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If external calibration using finger stick blood glucose meters is performed, then the glucose monitoring system can be calibrated, but the process is painful, costly, and prone to errors
Solution Approach 1:
The sensor system performs self-calibration by automatically detecting and analyzing sensor signals to determine calibration parameters without requiring external finger stick measurements. The microcontroller processes sensor data and applies calibration algorithms internally, eliminating the need for user-performed finger sticks while maintaining calibration accuracy.
Solution Approach 2:
The patent replaces the mechanical finger stick procedure with an electronic/self-contained calibration process. Instead of physically pricking the finger and applying blood to a test strip, the system uses electronic signal processing and machine learning algorithms to perform calibration automatically within the sensor device.
2Measurement precision
If external calibration using finger stick blood glucose meters is performed, then the glucose monitoring system can be calibrated, but the process is costly and time-consuming
Solution Approach 1:
The system performs calibration actions automatically and continuously in the background without requiring dedicated calibration time from the user. The microcontroller continuously processes sensor signals and updates calibration parameters as needed, eliminating the need for scheduled finger stick calibration sessions.
Solution Approach 2:
The calibration process is made continuous and ongoing rather than periodic and discrete. The system continuously monitors sensor signals and performs incremental calibration adjustments, maintaining accurate calibration throughout the sensor's operational life without interrupting the user's routine.
3Reliability
If traditional sensor systems are used, then glucose levels can be monitored, but the system cannot reliably distinguish between sensor failures and physiological changes
Solution Approach 1:
The system implements feedback mechanisms where the microcontroller continuously monitors sensor signals and compares them against expected physiological patterns. When anomalies are detected, the system can distinguish between sensor failures and actual physiological changes by analyzing the characteristics and context of the signals, providing reliable diagnostic information.
Solution Approach 2:
The patent utilizes changes in multiple sensor parameters and their relationships to diagnose system status. By monitoring changes in sensor current, impedance, and other electrical characteristics alongside glucose readings, the system can identify whether deviations are due to sensor failures or genuine physiological changes, enhancing diagnostic capability.
4Measurement precision
If frequent finger stick calibration is required, then calibration accuracy can be maintained, but user compliance and convenience deteriorate
Solution Approach 1:
The sensor system performs self-calibration automatically without requiring user intervention. The microcontroller independently processes sensor data, applies calibration algorithms, and maintains accurate calibration throughout operation, completely eliminating the burden of frequent finger stick procedures for the user.
Solution Approach 2:
The patent introduces an intermediary calibration mechanism that bridges the gap between sensor measurements and accurate glucose readings without requiring external finger sticks. The system uses internal reference measurements and algorithmic processing to maintain calibration accuracy, serving as an intermediary solution that preserves precision while improving ease of operation.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables retrospective calibration and enhanced reliability of glucose monitoring with reduced finger stick requirements, improving accuracy and autonomy of glucose monitoring systems.
Implementation Method 1
a sensor for producing signals indicative of a characteristic of a user
Implementation Method 2
decomposing the preprocessed Isig values using discrete wavelet decomposition
Implementation Method 3
using at least one machine learning model to calculate, by the microcontroller, a final sensor glucose (SG) value based on the Isig values and the discrete wavelet decomposition
Data Source
AI summary
A method for retrospective calibration of a glucose sensor uses stored values of measured working electrode current (Isig) to calculate a final sensor glucose (SG) value retrospectively. The Isig values may be preprocessed, discrete wavelet decomposition applied. At least one machine learning model, such as, e.g., Genetic Programing (GP) and Regression Decision Tree (DT), may be used to calculate SG values based on the Isig values and the discrete wavelet decomposition. Other inputs may include, e.g., counter electrode voltage (Vcntr) and Electrochemical Impedance Spectroscopy (EIS) data. A plurality of machine learning models may be used to generate respective SG values, which are then fused to generate a fused SG. Fused SG values may be filtered to smooth the data, and blanked if necessary.