Sparsely Sampled Analyte Sensor Noise Rejection by Combined Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional noise filtering methods for sparsely sampled analyte sensor data fail to distinguish between noise and rapid changes in analyte concentration, leading to over-filtering and lagging responses, especially in systems like continuous glucose monitors.
Innovation Solution
A method combining interpolation-based and extrapolation-based estimation techniques using least squares fit calculations to generate a combined estimate of analyte levels, which includes determining an interpolation-based estimate within a measurement window and extrapolation-based estimates outside the window, and then averaging these to produce a final estimate.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional noise filtering methods are applied to sparsely sampled analyte sensor data, then noise is reduced, but rapid changes in analyte concentration are attenuated and response time increases
Solution Approach 1:
The patent applies dynamics by making the filtering approach adaptive rather than static. The system dynamically switches between interpolation-based filtering (for stable conditions) and extrapolation-based filtering (for rapidly changing conditions) based on real-time analysis of data variability and rate of change, allowing the filter to respond appropriately to different signal characteristics without always attenuating rapid changes
Solution Approach 2:
The patent changes the fundamental parameter of the filtering method itself - switching between interpolation (using past and future points) and extrapolation (using only past or future points) based on signal characteristics. This parameter change allows the system to maintain noise reduction while preserving rapid excursions when they occur, resolving the contradiction between filtering effectiveness and response speed
2Device complexity
If sparsely sampled sensor data is used, then device complexity is reduced, but measurement accuracy deteriorates due to noise and inability to capture rapid changes
Solution Approach 1:
The patent applies preliminary action by pre-calculating both interpolation-based and extrapolation-based estimates from the sparsely sampled data, then combining them optimally. This preliminary processing of the limited data points allows the system to extract maximum information while maintaining accuracy, effectively compensating for the low sampling rate without requiring additional hardware complexity
Solution Approach 2:
The patent creates a composite estimation method by combining interpolation-based estimates and extrapolation-based estimates in a weighted manner. This composite approach leverages the strengths of both methods - interpolation for noise reduction and extrapolation for capturing rapid changes - thereby achieving high measurement precision from sparsely sampled data without increasing device complexity
3Stability of the object's composition
If aggressive noise filtering is applied, then signal smoothness is improved, but the sensor appears less responsive and lags behind reference measurements
Solution Approach 1:
The patent segments the filtering process into two distinct components: interpolation-based filtering that provides smoothness and noise reduction, and extrapolation-based filtering that preserves responsiveness to rapid changes. By segmenting the approach and applying each component appropriately based on signal characteristics, the system achieves both smoothness and reliability without the lag associated with aggressive single-method filtering
Data Source
AI summary
Systems, methods and apparatus are provided for rejecting noise from sparsely sampled analyte sensor data. Embodiments of the present disclosure include receiving a raw set of sensor data from an on-body device including an in vivo analyte sensor, determining an interpolation-based estimate of an analyte level over time based on the raw set of sensor data, determining an extrapolation-based estimate of the analyte level over time based on the raw set of sensor data, determining a combined estimate of the analyte level over time based on the interpolation-based estimate and the extrapolation-based estimate, and displaying a representation of the combined estimate of the analyte level over time on an output device. Numerous additional aspects are disclosed.