SpO2 Estimation Using Radial Basis Neural Network for Motion Artifact Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Pulse oximeters face challenges in accurately measuring oxygen saturation due to noise artifacts from motion and ambient light, leading to negative SpO2 bias, which degrades signal reliability.
Innovation Solution
A device and method using a radial basis neural network to combine multiple SpO2 estimates and signal quality metrics, adjusting for noise metrics to produce a corrected SpO2 estimate that minimizes motion-based errors, thereby improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If motion artifact is present during SpO2 measurement, then the measurement process can continue without interruption, but the SpO2 accuracy deteriorates with negative bias
Solution Approach 1:
The patent develops noise metrics that specifically quantify motion artifact effects and transforms this harmful factor into a useful correction mechanism. By measuring the noise characteristics introduced by motion, the system calculates compensation values that are added to the SpO2 estimate, converting the harmful motion artifact into a correctable parameter that improves measurement accuracy during motion.
Solution Approach 2:
The patent changes the parameter representation by introducing noise metrics as additional input parameters to the SpO2 calculation algorithm. Instead of using only the traditional photoplethysmography signals, the system incorporates quantified noise parameters that describe motion artifact characteristics, allowing the algorithm to adjust the SpO2 estimate based on the measured noise level and type.
2Reliability
If multiple SpO2 estimates are combined to improve accuracy, then the measurement reliability improves, but the algorithm complexity increases
Solution Approach 1:
The patent merges multiple internal SpO2 estimates along with their associated noise metrics into a single corrected SpO2 value. Rather than selecting one estimate from multiple candidates, the system combines them using a neural network that processes all inputs simultaneously, integrating the information from multiple measurement pathways and noise characterizations into one unified output.
Solution Approach 2:
The patent introduces a neural network as an intermediary processing layer between the raw SpO2 estimates and the final corrected value. This intermediary component learns the optimal weighting and combination strategy from training data, mediating the complex interactions between multiple estimates and noise metrics without requiring explicit programming of combination rules.
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
The solution effectively reduces noise bias in oxygen saturation estimates, providing more reliable and accurate readings even in the presence of motion artifacts, as demonstrated by improved correlation with non-motion data.
Implementation Method 1
process the plurality of oxygen saturation estimates and the plurality of signal quality metrics associated with the plurality of oxygen saturation estimates using a solved, computer-implemented radial basis neural network to produce a corrected oxygen saturation estimate
Data Source
AI summary
The present disclosure relates, according to some embodiments, to devices, systems, and methods for estimating a physiological parameter in the presence of noise. For example, the disclosure relates, in some embodiments, to devices, systems, and methods for assessing (e.g., estimating, measuring, calculating) oxygen saturation (SpO2). Methods of assessing SpO2 may include assessing a noise metric associated with motion artifact. In some embodiments, a percentage (e.g., an empirically determined percentage) of a noise metric may be simply added to the SpO2 estimate to produce a corrected SpO2 estimate. An oximetry algorithm may include, according to some embodiments, combining multiple internal SpO2 estimates and associated noise and/or signal quality metrics (e.g., using a radial basis neural network) to produce a modified (e.g., corrected) SpO2 estimate (e.g., rather than merely selecting the estimate from a finite number of candidates). A modified SpO2 estimate may include little or no movement-based error.