Heart Rate Variability Feature Extraction Using Neural Network Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for extracting heart rate variability feature values require long-term measurements, which are inconvenient and prone to signal deterioration due to user movement, necessitating the development of methods for reliable extraction using short-term measurements.
Innovation Solution
A method utilizing a pre-trained neural network to extract heart rate variability feature values from short-term biosignal data, where the neural network is trained on segmented data to output feature values corresponding to longer time periods, allowing for reliable extraction even in motion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If long-term measurement is performed to extract reliable heart rate variability feature values, then measurement reliability is improved, but user convenience deteriorates and signal quality may worsen due to movement
Solution Approach 1:
The neural network model is pre-trained using long-term measurement data to learn the relationship between short-term and long-term heart rate variability features. This preliminary training enables the model to predict reliable long-term feature values from short-term measurements, eliminating the need for users to undergo lengthy measurement sessions while maintaining measurement reliability.
2Reliability
If long-term measurement is performed to extract reliable heart rate variability feature values, then measurement reliability is improved, but signal quality deteriorates due to user movement
Solution Approach 1:
The neural network is pre-trained offline using high-quality long-term measurement data, enabling it to learn robust feature extraction patterns that are insensitive to movement artifacts. During actual use, the pre-trained model processes short-term measurements and predicts reliable long-term feature values without requiring the user to remain stationary for extended periods, thereby avoiding signal quality deterioration from movement.
3Ease of operation
If short-term measurement is used to improve user convenience, then ease of operation is improved, but measurement reliability deteriorates
Solution Approach 1:
The neural network model serves as an intermediary that bridges short-term measurements and long-term feature values. It takes short-term biosignal data as input and outputs predicted long-term heart rate variability feature values, enabling the system to provide reliable long-term analysis results based on short-term user-friendly measurements.
Solution Approach 2:
The patent transforms the measurement time parameter from the traditional long-term duration to a short-term duration by introducing the neural network prediction mechanism. The model compensates for the reduced measurement time by leveraging patterns learned during training, effectively changing the relationship between measurement duration and result reliability.
4Ease of operation
If short-term measurement is used to improve user convenience, then ease of operation is improved, but measurement precision deteriorates
Solution Approach 1:
The pre-trained neural network acts as an intermediary processing layer that enhances short-term measurement data to produce precise long-term feature estimates. The model compensates for the limited information in short-term measurements by applying knowledge learned from extensive training data, thereby maintaining measurement precision despite reduced measurement duration.
Data Source
AI summary
Disclosed is a method for extracting a heart rate variability (HRV) feature value performed by a computing device including one or more processors. The method includes acquiring first biosignal data measured during a first time period. The method includes outputting one or more heart rate variability feature values corresponding to a time period longer than the first time period by inputting the first biosignal data into a pre-trained neural network model.