Analyte Sensor Calibration With Dynamic Lag Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Analyte monitoring systems face challenges in accurately converting interstitial fluid analyte levels to blood analyte levels due to lag times, leading to inaccurate calibration and increased false alerts, particularly during critical periods like overnight usage.
Innovation Solution
Implementing a method that assigns weights to historical calibration points using exponential growth formulas and updates lag parameters based on time intervals, employing asymmetrical lag methodologies to enhance accuracy and reduce false alerts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If ISF analyte levels are used to calculate blood analyte levels, then continuous monitoring is achieved, but measurement precision deteriorates due to lag times
Solution Approach 1:
The patent dynamically adjusts the lag parameter based on the time since sensor insertion, recognizing that the lag between ISF and blood analyte levels changes over the sensor's lifespan. By updating the lag parameter periodically, the system adapts to changing physiological conditions and maintains measurement precision while continuing to use ISF measurements for blood analyte level calculation.
2Measurement precision
If calibration is performed frequently to maintain accuracy, then measurement precision improves, but device complexity increases due to multiple reference measurements
Solution Approach 1:
The system performs a warm-up period after sensor insertion during which no calibration is required. This preliminary action allows the sensor to stabilize and establishes initial operating conditions, delaying the need for calibration and reducing the frequency of reference measurements needed throughout the sensor's lifespan.
Solution Approach 2:
Calibration is performed periodically at scheduled intervals rather than continuously or at every measurement. The system updates the lag parameter and performs calibration only when needed based on predetermined time intervals, reducing the burden of frequent reference measurements while maintaining acceptable accuracy.
3Adaptability or versatility
If recent calibration points are given higher weight for current conditions, then adaptability improves, but reliability deteriorates due to over-fitting and false alerts
Solution Approach 1:
The weighting factor is dynamically adjusted based on the time since sensor insertion rather than being fixed. During the warm-up period, historical calibration points are weighted more heavily, while after the warm-up period, the system transitions to a different weighting strategy that balances recent and historical data, preventing over-fitting to transient conditions.
Solution Approach 2:
The patent employs asymmetrical weighting where the treatment of calibration points differs before and after the warm-up period. The weighting function creates an asymmetric profile that heavily favors historical data during warm-up, then transitions to a more balanced approach, rather than using a symmetric weighting scheme that would treat all periods equally.
4Adaptability or versatility
If lag parameters are updated continuously to reflect changing conditions, then adaptability improves, but measurement precision deteriorates due to instability in parameter estimation
Solution Approach 1:
The lag parameter is updated periodically at predetermined intervals rather than continuously with every measurement. This periodic update approach allows sufficient data to accumulate for stable parameter estimation while still adapting to changing conditions over the sensor's lifespan, balancing adaptability with measurement precision.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Disclosed are analyte monitoring systems and methods for calibrating an analyte sensor using one or more reference measurements. These systems and methods may include using a conversion function and first sensor data to calculate a first sensor analyte level, weighting a first reference analyte measurement (RM1) and one or more previous reference analyte measurements according to a weighted average cost function, updating the conversion function using the weighted RM1 and the one or more weighted previous reference analyte measurements as calibration points, and using the updated conversion function and second sensor data to calculate a second sensor analyte level. In some aspects, the systems and methods may include updating one or more of lag parameters used to calculate the sensor analyte levels.