Sleep Detection Using Sensor Fusion for Multi-Occupant Bed Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current sleep detection systems lack the accuracy and capability to differentiate between individuals in a shared bed, accurately track sleep stages, and identify respiratory events and snoring, which are crucial for understanding and improving sleep quality.
Innovation Solution
A sleep analytics system utilizing artificial intelligence and sensor fusion, including piezo force sensors, audio, and accelerometer data, to identify sleep stages, detect respiratory events, and attribute snoring to specific individuals, leveraging deep learning models for accurate classification and real-time analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple sensors are used to improve sleep detection accuracy, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent combines multiple sensor types (piezoelectric force sensors, accelerometers, audio sensors) into a unified sleep detection system that collects and processes data from all sensors simultaneously. This merging approach enables comprehensive sleep stage detection, respiratory event identification, and snoring attribution while maintaining manageable system complexity through integrated processing.
Solution Approach 2:
The piezoelectric force sensors serve multiple functions: detecting body movement, monitoring respiratory events, identifying snoring, and determining sleep stages. This multi-functionality reduces the need for separate specialized sensors, thereby improving measurement precision without proportionally increasing device complexity.
2Measurement precision
If sensor fusion with multiple data types is implemented to differentiate individuals in shared beds, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The system segments the analysis by creating individual profiles for each bed occupant based on their unique sensor data patterns. By dividing the complex task of identifying multiple individuals into separate analyzable profiles, the system achieves high identification accuracy while managing data processing complexity through structured segmentation.
Solution Approach 2:
The patent applies local quality by tailoring the analysis to each individual occupant's specific characteristics and patterns in the sensor data. This allows the system to differentiate between individuals in shared beds by focusing on their unique local data qualities rather than requiring a single complex unified analysis.
3Measurement precision
If deep learning models are used to identify sleep stages and respiratory events, then measurement precision is improved, but use of energy increases
Solution Approach 1:
The system applies partial action by using deep learning models selectively for specific analysis tasks (sleep stage identification, respiratory event detection) rather than continuously processing all data at maximum complexity. This approach achieves high measurement precision for critical functions while reducing overall energy consumption by applying intensive processing only where necessary.
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
The system provides high accuracy in sleep stage detection and identification of respiratory events and snoring, comparable to polysomnographic analysis, enabling users to gain insights into their sleep quality and receive alerts for potential health issues, such as sleep apnea.
Implementation Method 1
The detection system uses as its baseline force transmitted through a mattress from body movement due to breathing, heartbeat, micromovements, and larger scale movements. This piezo force sensor data is processed to capture all of the data
Data Source
AI summary
A method of detecting the presence of a person on a bed comprising, receiving movement data in pseudo-real-time, extracting features from the movement data for a time period in pseudo-real-time, determining whether the bed is occupied for the time period in pseudo-real-time, for each time period in which the bed is indicated as occupied, determining a user on a side of the bed, based on data from a plurality of time periods, and utilizing a sleep engine to determine sleep stages for the user. The method in one embodiment further comprising flagging when a majority of the plurality of time periods indicate that the sleep stage for the user is awake and providing a sleep recording to the user, showing the sleep stages for the user.