Sleep State Prediction Using Sheet Sensors and Phase Coherence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for activity state prediction, such as sleep polysomnography and actigraphy, face challenges in real-time accuracy and applicability to small animals due to complex sensor setups and detection limitations.
Innovation Solution
An activity state prediction system using a sheet-type sensor that calculates phase coherence from bio-vibration signals, including respiratory rate, heart rate, and heartbeat interval variation, employing machine learning with adaptive boosting to predict leaving-bed, awake, and sleeping states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sleep polysomnography with multiple sensors (EEG, ECG, EMG, EOG) is used to achieve accurate activity state prediction, then measurement precision is improved, but device complexity and ease of operation deteriorate due to complicated handling and multiple sensors attached to specified body sites
Solution Approach 1:
The patent extracts only the essential physiological signals needed for activity state prediction from the full polysomnography system. Instead of using all multiple sensors (EEG, ECG, EMG, EOG), it selectively uses body movement signals, respiratory rate, heart rate, and heartbeat interval variation obtained through simpler means, thereby reducing device complexity while maintaining prediction accuracy
Solution Approach 2:
The patent replaces the mechanical attachment of multiple sensors to specified body sites with a non-contact or minimal-contact measurement system. Body movement signals are obtained through acceleration sensors, and cardiovascular/respiratory signals are derived from photoplethysmography or similar optical methods, eliminating the need for complex sensor attachments
2Device complexity
If actigraphy technique is used to simplify the sensor setup, then device complexity is reduced, but reliability deteriorates because it cannot deliver real-time prediction results as it predicts current state based on previous and future data
Solution Approach 1:
The patent performs preliminary processing of physiological signals to extract features that can predict current state without requiring future data. By pre-extracting body movement signals, respiratory rate, heart rate, and heartbeat interval variation, and computing phase coherence from these signals, the system can make real-time predictions based only on past and present data, eliminating the need for future data while maintaining reliability
Solution Approach 2:
The patent implements a feedback mechanism where the predicted activity state is continuously updated based on incoming physiological signals. The phase coherence computation and machine learning model process streaming data in real-time, providing continuous feedback that allows the system to adapt and maintain accurate real-time predictions without requiring future data
3Device complexity
If tube-shaped body movement detection means is used to simplify the detection structure, then device complexity is reduced, but measurement precision deteriorates when the subject animal is smaller than the size of the mat, causing body movement to be undetected due to the animal being outside the detection means
Solution Approach 1:
The patent uses a sheet-type sensor that can detect body movement across its entire surface area, making it universally applicable to animals of various sizes. Unlike tube-shaped detection means that only detect movement within a limited volume, the sheet-type sensor covers the entire mat surface, allowing it to detect body movement of small animals anywhere on the mat while maintaining simple structure
Solution Approach 2:
The patent transitions from three-dimensional tube-shaped detection volume to two-dimensional sheet-type sensor surface. This dimensional change allows the detection means to cover a larger area with simpler structure, enabling detection of body movement across the entire mat surface regardless of animal size, thereby improving measurement precision without increasing device complexity
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables real-time, accurate prediction of activity states with high agreement and low false positives, reducing caregiver burden and enabling early detection of developmental issues in infants and newborns.
Implementation Method 1
a sensor part configured to detect bio-vibration signals of an animal
Data Source
AI summary
The present invention delivers an activity state prediction system predicts leaving-bed/awake/sleeping state in a real-time and accurate manner. An activity state prediction model is established using information on leaving-bed/awake/sleeping state based on sleep polysomnography data as teaching data and body movement signal, respiratory rate, heart rate, and phase coherence acquired simultaneously with the teaching data as input data and then predicts leaving-bed/awake/sleeping state in response to entry of data on the body movement signal, respiratory rate, heart rate, and phase coherence obtained from the bio-vibration signals detected by a sensor.