Wearable Biosignal Monitoring Using Segmented AI Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biosignal monitoring technologies face challenges in real-time analysis due to resource limitations in wearable devices and high data transmission requirements between devices and servers, leading to delayed analysis results.
Innovation Solution
Implementing a primary analysis using a lightweight model in the wearable device and a secondary analysis using an advanced model on a server, with data reduction by removing low-frequency noise to minimize data transmission and processing burdens.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If an advanced AI analysis model is implemented in the wearable device for real-time biosignal analysis, then analysis accuracy is improved, but device resource consumption exceeds available capacity
Solution Approach 1:
The patent segments the AI analysis process into two distinct parts: a lightweight primary analysis model executed on the wearable device for real-time monitoring, and an advanced secondary analysis model executed on the server for comprehensive diagnosis. This segmentation allows each component to operate within its resource constraints while achieving overall high accuracy.
Solution Approach 2:
The patent introduces a server as an intermediary between the wearable device and the advanced AI analysis. The server receives biosignal data from the device, executes the resource-intensive advanced analysis model, and returns results to the device. This intermediary approach enables complex analysis without burdening the wearable device's limited resources.
2Measurement precision
If all biosignal data is transmitted to the server for analysis, then analysis completeness is improved, but data transmission time and communication burden increase
Solution Approach 1:
The patent extracts and transmits only the essential primary analysis results and critical biosignal features to the server, rather than transmitting all raw biosignal data. This extraction approach maintains analysis completeness while significantly reducing data transmission volume and time.
Solution Approach 2:
The patent performs preliminary primary analysis on the wearable device before data transmission. This preliminary processing filters and pre-processes the biosignal data, extracting key features and abnormalities that need server-side attention, thereby reducing the amount of data requiring transmission while ensuring no critical information is lost.
3Productivity
If a lightweight analysis model is used in the wearable device, then processing speed is improved, but analysis accuracy deteriorates
Solution Approach 1:
The patent segments the analysis into two stages with different accuracy-speed tradeoffs: the primary analysis model prioritizes speed for real-time monitoring, while the secondary analysis model prioritizes accuracy for final diagnosis. Together, they achieve both real-time responsiveness and high accuracy.
Solution Approach 2:
The patent implements a feedback mechanism where the primary analysis results are sent to the server, which performs secondary analysis and returns refined results to the device. This feedback loop allows the lightweight device model to benefit from the enhanced accuracy of the server-side advanced model while maintaining real-time operation.
Data Source
AI summary
A method for monitoring biosignals using a wearable device is provided. The method includes the steps of: acquiring information on a result of performing a primary analysis of a biosignal measured by the device, using a primary analysis model trained to perform a primary analysis for detecting abnormal events from biosignals, and acquiring, from the biosignal, a partial biosignal associated with the result of performing the primary analysis; and performing a secondary analysis of the partial biosignal with reference to the information on the result of performing the primary analysis, using a secondary analysis model trained to perform a secondary analysis for detecting abnormal events from biosignals, wherein the primary analysis model is a relatively light-weighted model compared to the secondary analysis model.