Multi-Source PPG Preprocessing for Accurate Blood Pressure Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing blood pressure prediction methods using photoplethysmograph (PPG) signals are prone to distortion due to sensor sensitivity, physiological status, and environmental interference, leading to inaccurate predictions when acquired directly or indirectly from videos.
Innovation Solution
A method and device utilizing multiple data sources with signal filtering and shaping techniques for direct PPG signals and video quality detection for indirect signals, followed by normalization to generate a uniform standard PPG data sequence, employing two convolutional neural network (CNN) models for prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If PPG signal is acquired directly through sensor, then signal acquisition is straightforward, but signal distortion occurs due to sensor sensitivity and environmental interference
Solution Approach 1:
The patent introduces multiple preprocessing modules as intermediaries between the sensor and the prediction model. These include baseline drift removal modules, normalized filtering modules, and video quality detection modules that act as mediators to clean and standardize the PPG signal, thereby resolving the contradiction between easy acquisition and accurate measurement
Solution Approach 2:
The patent applies parameter changes through normalized filtering that adjusts signal parameters (amplitude, baseline) to standard ranges. The preprocessing transforms raw signals with varying parameters into standardized signals, enabling accurate prediction while maintaining operational simplicity
2Adaptability or versatility
If PPG signal is extracted from video, then non-contact measurement is achieved, but signal distortion occurs due to light intensity variations
Solution Approach 1:
The patent introduces video quality detection and normalized signal conversion modules as intermediaries between video capture and PPG extraction. These modules detect video quality parameters and convert varied light intensity signals into standardized PPG signals, resolving the contradiction between measurement versatility and precision
Solution Approach 2:
The patent performs preliminary video quality detection and signal normalization before PPG extraction. By pre-processing the video signal to detect quality issues and normalize intensity variations beforehand, the system ensures accurate PPG extraction while maintaining the versatility of non-contact measurement
3Adaptability or versatility
If multiple preprocessing methods are applied to different data sources, then compatibility with various data sources is improved, but system complexity increases
Solution Approach 1:
The patent segments the preprocessing system into distinct modules tailored to different data sources: baseline drift removal for direct PPG signals, normalized filtering for PPG signals, and video quality detection with normalized signal conversion for video-derived PPG signals. This segmentation enables handling of multiple data sources while keeping each module's complexity manageable
Solution Approach 2:
The patent creates a universal preprocessing framework that handles multiple data source types through standardized output formats. Despite different preprocessing methods for different sources, all modules produce standardized PPG signals that feed into the same prediction model, achieving versatility without proportionally increasing overall system complexity
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
Improves the processing capacity and compatibility of PPG signal data sources, enabling accurate blood pressure prediction by addressing signal distortions and enhancing the compatibility with various data sources.
Implementation Method 1
a PPG signal reflecting the light absorption capacity of blood also changes periodically
Implementation Method 2
identifying the light intensity of a specific light source with a light sensor
Data Source
Figure 1~2
Figure 3~4
AI summary
The embodiments of the invention relate to a blood pressure prediction method and device using multiple data sources. The method comprising: acquiring a data source identifier and original data (1); performing a data preprocessing operation on the original data according to the data source identifier(2); when the data source identifier is a first-class PPG original signal identifier, performing normalized filtering on a first-class PPG original signal to generate a standard PPG data sequence(21); when the data source identifier is a second-class PPG original signal identifier, performing baseline drift removal and normalized filtering on a second-class PPG original signal to generate a standard PPG data sequence(22); when the data source identifier is a third-class PPG video identifier, performing video quality detection and normalized signal conversion on third-class video data to generate a standard PPG data sequence(23); obtaining a CNN model identifier(3); and selecting a corresponding CNN model to predict blood pressure on the standard PPG data sequence according to the data source identifier and/or the CNN model identifier(4).