In-Ear Sensor Signal Separation for Sleep Staging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current sleep stage monitoring technologies are cumbersome, expensive, and lack accuracy, particularly for whole-night sleep staging, as they require multiple sensors and are uncomfortable for users, limiting their effectiveness in diagnosing and evaluating sleep quality.
Innovation Solution
A computer-implemented method and in-ear sensing device that separates EEG, EOG, and EMG signals from a single-channel in-ear signal using a spectral template matrix and activation matrix, optimized through a divergence function, allowing for accurate sleep stage classification with a lightweight and inexpensive wearable system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional polysomnography (PSG) with multiple sensors is used for sleep stage monitoring, then measurement precision is improved, but device complexity and ease of operation deteriorate due to cumbersome sensor placement and multiple electrodes
Solution Approach 1:
The patent segments the complex multi-sensor PSG system into a simplified single-channel in-ear sensing system. By placing sensors in the ear canal, it captures EEG, EOG, and EMG signals through a single location rather than requiring multiple electrodes distributed across the head and face, thereby reducing device complexity while maintaining measurement precision through signal separation algorithms
Solution Approach 2:
The patent merges multiple measurement functions (EEG for brain activity, EOG for eye movements, EMG for muscle tone) into a single in-ear sensing channel. The ear canal location serves as a convergence point where all three signal types can be captured simultaneously through one sensor placement, eliminating the need for separate electrode placements for each measurement modality
2Measurement precision
If traditional polysomnography (PSG) with multiple sensors is used for sleep stage monitoring, then measurement precision is improved, but ease of operation worsens due to professional installation requirements and lead placement complexity
Solution Approach 1:
The patent segments the complex multi-step electrode placement process into a single in-ear insertion operation. Instead of requiring professionals to place multiple electrodes at specific locations on the head and face according to standardized protocols, the system simplifies installation to inserting a single in-ear device, making it suitable for consumer use without professional intervention
Solution Approach 2:
The in-ear sensing device is designed for self-service installation by the user without requiring professional technicians. The device can be easily inserted and removed by the user themselves, eliminating the need for professional installation and making the system accessible for home use and long-term monitoring
3Ease of operation
If in-ear sensing with single channel is used, then ease of operation is improved, but measurement precision deteriorates due to signal mixing of EEG, EOG, and EMG
Solution Approach 1:
The patent applies preliminary action by pre-computing spectral templates for EEG, EOG, and EMG signals during a training phase using data from traditional PSG. These pre-established templates are then used to guide the separation of mixed signals in real-time applications, enabling accurate decomposition without requiring complex real-time computation or multiple sensors
Solution Approach 2:
The patent implements feedback through an iterative optimization process that uses divergence minimization to refine the separation of mixed signals. The system continuously adjusts the decomposition of the single-channel signal by comparing it against the pre-established spectral templates and minimizing the divergence between the separated components and expected signal characteristics, thereby improving measurement precision
Data Source
AI summary
The present invention provides a light-weight wearable sensor that can capture electroencephalogram (EEG or brain signals), electromyography (EMG or muscle signal), and electrooculography (EOG or eye movement signal) using a pair of modified off-the-shelf earplugs. The present invention further provides a supervised non-negative matrix factorization learning algorithm to analyze and extract these signals from the mixed signal collected by the sensor. The present invention further provides an autonomous and whole-night sleep staging system utilizing the sensor's outputs.