Physiological Signal Filtering With Window Error Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for real-time physiological signal monitoring, such as respiratory rate, often require specialized, expensive devices and are susceptible to noise, making them impractical for everyday use and inaccurate in dynamic environments.
Innovation Solution
A system and method using a smartwatch with an Inertial Measurement Unit (IMU) that employs a filter machine learning model to predict errors in physiological signal predictions, segmenting data into temporal windows, and discarding unreliable data, while using an extraction model like a Convolutional Neural Network (CNN) to accurately extract signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If specialized invasive devices are used for physiological signal monitoring, then measurement precision is improved, but device complexity and ease of operation deteriorate
Solution Approach 1:
The patent extracts the essential function of physiological signal monitoring from complex specialized devices and implements it using standard smartphone sensors (accelerometer, gyroscope, barometer). The extraction machine learning model isolates the core signal processing capability, separating it from the cumbersome hardware requirements of traditional monitoring devices.
Solution Approach 2:
The patent creates a functional copy of specialized monitoring device capabilities using readily available smartphone sensors. Instead of requiring dedicated medical-grade equipment, the system replicates monitoring functionality through machine learning models that process data from common smartphone components, making the technology accessible without specialized hardware.
2Measurement precision
If motion artifact filtering is applied to improve signal accuracy, then measurement precision is improved, but computational burden increases
Solution Approach 1:
The filter machine learning model is trained in advance on labeled data containing both clean signals and motion-corrupted signals. This preliminary training enables the model to automatically distinguish between actual physiological changes and motion artifacts during runtime, eliminating the need for complex real-time filtering algorithms and reducing computational burden during deployment.
Solution Approach 2:
The patent introduces an intermediary filter machine learning model that sits between the raw sensor data and the extraction model. This intermediary component predicts whether input data contains motion artifacts and selectively filters or flags problematic data, simplifying the overall processing pipeline while maintaining signal accuracy.
3Reliability
If continuous monitoring is performed to improve reliability, then reliability is improved, but energy consumption increases
Solution Approach 1:
The system implements periodic monitoring at strategically determined intervals rather than truly continuous monitoring. The filter model identifies periods when motion artifacts are present and temporarily suspends or reduces monitoring frequency during those periods, while maintaining reliable monitoring during stable periods. This periodic approach maintains reliability while significantly reducing energy consumption compared to continuous monitoring.
Solution Approach 2:
The filter machine learning model provides feedback about data quality and motion artifact presence to the monitoring system. Based on this feedback, the system dynamically adjusts monitoring intensity and frequency - intensifying monitoring when data quality is high and reducing it when motion artifacts are detected - thereby maintaining reliability while optimizing energy usage based on real-time conditions.
Data Source
AI summary
Systems and methods for filtering time-varying data for filtering and extracting a predicted physiological signal. A method including: segmenting the time-varying data into temporal windows; using a trained filter machine learning model, predicting an error for each prediction of the physiological signal for each window of time-varying data, the filter machine learning model trained using physiological signal predictions based on training time-varying data and known values of the physiological signal for the training time-varying data; discarding each window of time-varying data when the predicted error for such window is greater than a threshold; and where the window of time-varying data is not discarded, outputting at least one of the window of time-varying data and the predicted error for each prediction of the physiological signal.