PPG Signal Quality Scoring for Wearable Biomarker Reliability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing PPG signal quality measurement methods are challenged by confounding factors such as motion, wear, and physiological variations, leading to inaccurate digital measures and a lack of generalizability across different wearable devices and anatomical locations, with limited ability to provide nuanced quality assessments.
Innovation Solution
A method utilizing self-normalized features like Shannon entropy, spectral SNR, kurtosis, and pulse morphology similarity, combined with motion-related features, to predict a continuous quality score and categorize quality issues, trained on real-world datasets to enhance generalizability across devices and locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple signal quality metrics are used to measure all aspects of signal quality, then measurement precision is improved, but device complexity increases due to different units and scales making interpretation challenging
Solution Approach 1:
The patent combines multiple signal quality metrics (SNR, entropy, kurtosis, pulse morphology similarity) into a unified quality score through a machine learning model. This merging approach maintains measurement precision by incorporating multiple aspects of signal quality while reducing device complexity by providing a single interpretable output rather than multiple metrics with different units and scales.
Solution Approach 2:
The machine learning model acts as an intermediary that processes multiple signal quality metrics and transforms them into a unified quality score. This intermediary approach allows the system to benefit from comprehensive signal quality measurement while simplifying the output for practical use, resolving the contradiction between precision and complexity.
2Ease of operation
If hand-engineered rules for signal quality are defined, then ease of operation is improved, but adaptability deteriorates as they fail to generalize across different downstream tasks
Solution Approach 1:
The machine learning model performs self-service by automatically learning optimal feature combinations and weighting from training data, eliminating the need for hand-engineered rules. This approach maintains ease of operation through automated quality assessment while significantly improving adaptability across different downstream tasks through the model's ability to generalize from diverse training examples.
Solution Approach 2:
The system transitions from fixed hand-engineered rules to a flexible machine learning model that can adapt parameters based on training data. This parameter change enables the system to maintain ease of operation through automated decision-making while achieving broad adaptability across different tasks and signal quality scenarios.
3Ease of manufacture
If PPG signal quality is measured for specific tasks, then ease of manufacture is improved, but adaptability deteriorates as the list of confounders is task-dependent and cannot generalize
Solution Approach 1:
The machine learning model achieves universality by being trained on diverse real-world data from multiple devices and anatomical locations. This enables the same model to accurately assess signal quality across different tasks, devices, and locations, resolving the contradiction between ease of manufacture through standardized measurement and adaptability through generalization.
Solution Approach 2:
The system performs preliminary action by pre-training the machine learning model on comprehensive real-world data before deployment. This preliminary training enables the model to handle various confounders and signal quality issues across different tasks, devices, and locations without requiring task-specific customization, thus achieving both ease of manufacture and broad adaptability.
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
Provides a granular quality score and type-specific issue identification, improving the reliability of digital biomarker measurements and compliance monitoring by accounting for diverse real-world artifacts.
Implementation Method 1
Blood volume measurements, such as photoplethysmography (PPG), are sensing modalities commonly used in wearable devices
Data Source
AI summary
Systems and methods for analyzing a blood volume signal are described. In an example, the method comprises extracting self-normalized features from blood volume signals within a signal window; generating a signal quality prediction based on the self-normalized features to provide a quality prediction score, wherein the quality prediction score is between an upper bound and a lower bound; and generating a health metric score based on the blood volume signals in the signal window if the quality prediction score is above a predetermined threshold. In an example, the blood volume signals are photoplethysmography (PPG) signals. In an example, the method includes identifying a signal quality issue if the quality prediction score is below the predetermined threshold.