Wearable Sensor Measurement Confidence via Signal Quality Metrics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Wearable devices face challenges in accurately determining measurement confidence for biometric data due to hardware and software limitations, leading to inappropriate data processing and reduced confidence in feedback accuracy, especially in real-time non-parametric data and when determining if the device is being worn by a user.
Innovation Solution
Implementing a wearable device with sensors that collect and process biometric data, using frequency transformation to create non-parametric data sets for probabilistic modeling, estimating signal quality, and determining measurement confidence through signal quality metrics, which helps validate the accuracy of biometric data and confirm if the device is being worn.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If wearable devices use multiple biometric sensors to collect and process data, then the quantity and speed of feedback is improved, but the accuracy and confidence in measurement data deteriorates due to hardware and software limitations
Solution Approach 1:
The system performs preliminary actions by collecting reference biometric data during a calibration period when the device is known to be worn correctly. This reference data is stored and used later to compare against new measurements, enabling the system to validate measurement confidence in real-time without sacrificing feedback speed.
Solution Approach 2:
The system implements feedback by continuously comparing new biometric measurements against the reference data to calculate measurement confidence scores. This feedback mechanism allows the device to identify and filter low-quality measurements while maintaining high feedback speed through efficient computational algorithms.
2Productivity
If the device processes all collected data as biometric measurement data, then the productivity is improved, but the reliability deteriorates due to inappropriate data being processed
Solution Approach 1:
The system extracts and separates appropriate biometric measurement data from inappropriate data using measurement confidence analysis. By calculating confidence scores based on reference data comparison, the system identifies and extracts only the reliable measurements for further processing, while filtering out noise and artifacts.
Solution Approach 2:
The system changes parameters by introducing measurement confidence scores as an additional parameter to evaluate data quality. This allows the system to dynamically adjust which data points are processed based on their confidence levels, maintaining high throughput while ensuring reliability through parameter-based filtering.
3Measurement precision
If the device determines measurement confidence using reference data and signal quality metrics, then the measurement precision is improved, but the device complexity increases
Solution Approach 1:
The system segments the complexity by dividing the measurement validation process into distinct modules: reference data collection, signal quality metric calculation, confidence score computation, and data filtering. This segmentation allows each component to be optimized independently and processed efficiently in a pipeline architecture.
Solution Approach 2:
The system applies partial action by calculating measurement confidence selectively based on signal quality thresholds. Instead of performing complex confidence analysis on every single data point, the system first filters data using signal quality metrics and only performs detailed confidence analysis on data that passes initial quality checks, reducing overall computational complexity.
4Reliability
If the device uses voltage threshold checks and signal quality metrics to determine wear status, then the reliability of wear detection is improved, but the measurement time increases
Solution Approach 1:
The system uses periodic action by implementing a multi-stage verification process with voltage threshold checks performed at regular intervals during the calibration period. This periodic sampling approach allows the system to accumulate sufficient evidence for wear status determination without requiring continuous analysis, balancing reliability with time efficiency.
Data Source
AI summary
Systems and methods determining measurement confidence for data collected from sensors of a wearable device are herein disclosed. In an implementation, a confidence measurement that a wearable device is worn by a user can be determined by determining that the wearable device is not in motion, comparing sample voltages collected using a light emitter to thresholds indicative of a surface on which the light is being reflected, calculating a signal quality metric using data collected using a pulse oximeter, and comparing the signal quality metric to thresholds indicative of typical biometric data measurements. Other implementations for confidence measurement can include frequency transforming signal data stored in a buffer, performing probabilistic modelling on the frequency transformed data, and determining a confidence measurement using a signal quality estimation based on the modelled data.