Autoregressive Respiratory Rate Extraction via Particle Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for extracting respiratory rates from pulse oximeter signals, particularly autoregressive (AR) models, face challenges in accuracy at high breathing rates and low Signal-to-Noise Ratios (SNR), and require improved models to reliably extract respiratory rates across varying conditions.
Innovation Solution
The implementation of an autoregressive model using projection onto linearly independent non-orthogonal bases with optimal parameter search (OPS) and particle filtering (PF) techniques, which factorize AR parameters into multiple pole terms and select the pole with the highest magnitude to represent respiratory rate, enhancing accuracy and robustness across different SNR levels and breathing rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional autoregressive (AR) model methods are used for respiratory rate extraction, then the method is computationally efficient and works with short data records, but the accuracy deteriorates at high breathing rates and low signal-to-noise ratios
Solution Approach 1:
The patent transforms the fixed-parameter AR model into an adaptive system by introducing time-varying parameters through particle filtering. The model order, pole locations, and signal characteristics are dynamically adjusted based on instantaneous signal conditions, enabling accurate tracking of respiratory rates across varying breathing patterns and noise levels without requiring manual reconfiguration.
Solution Approach 2:
The invention transitions from a static AR model to a dynamic particle filtering framework where model parameters evolve over time. Multiple particle hypotheses represent different possible respiratory rate states, and these particles are continuously updated based on new signal arrivals, allowing the system to adapt to changing breathing rates and noise conditions in real-time.
2Measurement precision
If time-frequency spectral techniques (CWT, VFCDM) are used to extract respiratory rates, then accuracy is improved for low to moderate breathing rates, but reliability deteriorates with increased respiratory rates
Solution Approach 1:
The patent segments the continuous signal analysis into discrete time windows, with each window processed independently by the particle filter to estimate local respiratory rate characteristics. This segmentation allows the model to capture transient changes in breathing patterns while maintaining computational tractability and avoiding the accumulation of errors over long recording periods.
Solution Approach 2:
The particle filtering algorithm is applied periodically to successive segments of the pulse oximeter signal, with each application producing an updated estimate of respiratory rate. This periodic re-estimation allows the system to track time-varying respiratory patterns and adapt to changing physiological conditions, improving reliability at high breathing rates where continuous methods struggle.
Data Source
AI summary
Accurate AR method for extracting respiratory rates directly from a pulse oximeter and accurate methods of extracting respiratory rates directly from a pulse oximeter under low signal-to-noise ratio (SNR) conditions.