Sleep Stage Classification Using Heart Rate Variability Parameters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sleep stage determination methods using heart rate signals are limited in accuracy, particularly in distinguishing between the first and second non-REM sleep stages (N1 and N2), and fail to accurately determine all five sleep stages, with a maximum precision of about 0.69.
Innovation Solution
A sleep determination apparatus and method utilizing heart rate variability parameters, including very low frequency (VLF), low frequency (LF), and high frequency (HF) components, along with mean heartbeat interval (RRI) and LF/HF ratio, employ machine learning techniques such as random forests and recurrent neural networks to distinguish between all five sleep stages, including REM, N1, N2, N3, and N4, with improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If PSG device with multiple electrodes is used to determine sleep stage, then measurement accuracy is improved, but device complexity and operational difficulty increase
Solution Approach 1:
The patent extracts and utilizes specific characteristic components (heart rate variability parameters: VLF, LF, HF components and LF/HF ratio) from the complex PSG measurement system. By focusing on these extracted parameters that are indicative of sleep stages, the system achieves accurate sleep stage determination without requiring the full PSG equipment with multiple electrodes
Solution Approach 2:
The patent introduces machine learning models (random forest and recurrent neural network) as intermediary processing layers between the simple heart rate signal input and sleep stage determination output. These intermediary algorithms process the heart rate variability parameters and transform them into accurate sleep stage classifications, bridging the gap between simple measurement and complex determination
2Measurement precision
If PSG test is conducted by skilled expert, then sleep stage determination accuracy is improved, but operational complexity and cost increase
Solution Approach 1:
The patent implements a self-service system where the machine learning model automatically performs sleep stage determination without requiring skilled experts. The system autonomously processes heart rate variability parameters and generates sleep stage classifications, making the determination process accessible to general users without specialized training
Solution Approach 2:
The patent replaces the mechanical system of expert human analysis with an automated computational system. The random forest and recurrent neural network algorithms substitute for the expert's analytical capabilities, performing sleep stage determination through computational processing rather than human expertise
3Ease of operation
If simple heart rate signal is used for sleep stage determination, then ease of operation is improved, but measurement precision deteriorates
Solution Approach 1:
The patent performs preliminary processing of the heart rate signal to extract meaningful features before sleep stage determination. By pre-calculating heart rate variability parameters (VLF, LF, HF components and LF/HF ratio) from the raw heart rate signal, the system prepares optimized input data that enhances the accuracy of subsequent sleep stage classification
Solution Approach 2:
The patent transforms the simple heart rate signal into multiple derived parameters (VLF, LF, HF components and LF/HF ratio) that contain richer information about sleep stages. This parameter transformation process converts a single simple measurement into multiple informative features that improve determination accuracy
4Ease of operation
If only four sleep stages are determined, then ease of operation is improved, but measurement precision deteriorates
Solution Approach 1:
The patent segments the sleep stages into five distinct categories (wake, N1, N2, N3, and REM) rather than combining them into four groups. This segmentation allows for more precise differentiation between sleep stages, particularly separating N1 and N2 as distinct stages, thereby improving measurement precision while maintaining computational feasibility through the machine learning models
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An object of the present invention is to provide a sleep determination apparatus, a sleep determination method, and a sleep determination program that provide an accurate and detailed determination of a sleep stage. A sleep determination apparatus of the present invention comprising: a heart rate data obtaining means for obtaining, based on data regarding a heart rate of a body of a user, a very low frequency component (VLF), a ratio of a low frequency component (LF) to a high frequency component (HF), a mean heartbeat interval (RRI), and a standard deviation of each heartbeat interval (RRI) of the heart rate as heart rate variability parameters indicating a heart rate state; and a sleep state determining means for determining which of the first non-REM sleep stage or the second non-REM sleep stage the heart rate variability parameters of the user indicate, in accordance with a first determination condition that is set based on a correlation between the heart rate variability parameters and the first non-REM sleep stage and a second determination condition that is set based on a correlation between the heart rate variability parameters and the second non-REM sleep stage.