Bayesian Peak Selection for Motion-Robust PPG Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wearable devices fail to accurately detect and analyze physiological data due to motion artifacts and other noise sources, which can obscure desired physiological information that may be included in or derived from the waveform.
Innovation Solution
A method and system that utilize a priori knowledge, such as predetermined data from sensors distinct from the physiological sensor, to determine and assign probabilities of validity to each peak in a physiological waveform, selecting a subset of peaks based on these probabilities to generate a more accurate physiological assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If PPG sensor is used to measure blood flow information, then physiological assessment can be generated, but motion artifacts and noise sources obscure the signal and reduce measurement precision
Solution Approach 1:
The patent introduces an intermediary signal processing system that uses Bayesian peak selection and probability calculations to filter out motion artifacts from PPG signals. The system computes probabilities for detected peaks based on predetermined data and selects only those peaks above a threshold, effectively separating useful physiological information from harmful motion artifacts.
Solution Approach 2:
The patent replaces traditional mechanical signal filtering methods with a computational approach using Bayesian statistics and probability theory. Instead of relying solely on analog filters, the system uses digital signal processing to identify and remove artifacts, improving precision while maintaining adaptability to different motion conditions.
2Measurement precision
If traditional peak detection algorithms are used, then processing speed is maintained, but measurement precision is reduced due to inability to filter noise
Solution Approach 1:
The patent performs preliminary actions by pre-calculating probability thresholds and using predetermined data from multiple sensors before final peak selection. This allows the system to quickly identify and reject erroneous peaks without requiring complex real-time computations, maintaining processing speed while improving accuracy.
Solution Approach 2:
The patent extracts only the essential information needed for peak validation from the complex signal data. By focusing calculations on probability computations rather than processing every signal component, the system achieves high precision peak detection while maintaining efficient processing speeds suitable for real-time applications.
3Measurement precision
If multiple sensors are integrated to improve measurement precision, then physiological assessment accuracy increases, but device complexity increases
Solution Approach 1:
The patent designs a multi-functional signal processing system that handles multiple sensor types through a unified Bayesian framework. The same probability computation and peak selection algorithms work across different sensor configurations, allowing the system to accommodate various sensor combinations without requiring separate processing logic for each, thus managing complexity while maintaining precision.
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
This approach effectively removes or filters out peaks that may be attributed to noise or are otherwise inaccurate with respect to the desired physiological information contained in the waveform, thereby increasing the accuracy of physiological assessments.
Implementation Method 1
Photoplethysmography (PPG) is based upon shining light into the human body and measuring how the scattered light intensity changes with each pulse of blood flow
Data Source
Figure 1
Figure 2~4
Figure 5A
AI summary
A wearable device includes at least one physiological sensor configured to detect and/or measure physiological information from a subject over a period of time when the wearable device is worn by the subject, and a processor coupled to the sensor. The processor is configured to detect respective peaks in a physiological waveform representing the physiological information, compute probabilities for the respective peaks based on predetermined data indicative of one or more conditions, select a subset of the respective peaks based on the probabilities thereof as representing more accurate physiological information for the subject, and generate a physiological assessment of the subject based on the subset of the respective peaks that was selected. Related signal processing devices, methods of operation, and computer program products are also discussed.