Blood Pressure Prediction Using Physiological Signal Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing non-invasive blood pressure measurement systems face challenges in accuracy due to factors like cuff size, elastic effects, and subject posture, and struggle with high-dimensional physiological data that includes outlier samples.
Innovation Solution
A system and method that utilize a processor to receive requests for blood pressure determination, obtain data including heart activity and personal information, extract target features, and apply a prediction model followed by an optimization model to accurately predict blood pressure, while reducing data dimensions without removing outlier samples.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If frequent blood pressure measurement is performed using sphygmomanometer, then blood pressure data can be obtained continuously, but subject discomfort increases due to frequent cuff inflation oppressing blood vessels
Solution Approach 1:
The patent replaces the mechanical sphygmomanometer measurement system with a computational model that uses physiological signals (PPG, ECG, respiratory signals) to predict blood pressure. This substitution eliminates the need for mechanical cuff inflation, allowing continuous measurement without physical discomfort to the subject.
Solution Approach 2:
The patent introduces physiological signals (photoplethysmogram, electrocardiogram, respiratory signals) as intermediary data to infer blood pressure indirectly. Instead of directly measuring blood pressure through mechanical means, the system uses these intermediate physiological parameters as proxies to calculate blood pressure values continuously.
2Reliability
If high-dimensional physiological characteristics data from multiple subjects is used to build prediction model, then model comprehensiveness improves, but model building difficulty increases due to data volume and outlier samples
Solution Approach 1:
The patent extracts and removes outlier samples from the high-dimensional physiological data before model training. By identifying and eliminating abnormal or erroneous data points, the system reduces data complexity and improves model training efficiency while preserving the comprehensive nature of the remaining dataset.
Solution Approach 2:
The patent performs preliminary data preprocessing including outlier removal, normalization, and feature selection before building the prediction model. This preliminary action on the data reduces its dimensionality and complexity, making the subsequent model training process more efficient and manageable.
Data Source
AI summary
A method implemented on a computing device having at least one processor, storage, and a communication platform connected to a network for determining blood pressure includes: receiving a request to determine a blood pressure of a first subject from a terminal, obtaining data relating to the first subject, the data relating to the first subject including data relating to heart activity of the first subject and personal information relating to the first subject, extracting target features relating to the first subject from the data relating to the first subject, determining a preliminary blood pressure of the first subject using a prediction model based on the target features relating to the first subject, determining a predicted blood pressure of the first subject using an optimization model based on the preliminary blood pressure and sending the predicted blood pressure of the first subject to the terminal in response to the request.