Non-contact AHI Estimation via Bio-vibration Feature Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating sleep apnea-hypopnea index (AHI) are either invasive or require complex equipment, necessitating a non-contact, unrestrained, and simplified approach using bio-vibration signals.
Innovation Solution
An apparatus that extracts and processes bio-vibration signals to create feature images from parameters like respiratory rate, heart rate, body movement, and phase coherence, which are then input into a machine learning model for AHI estimation, eliminating the need for pneumotachographs and pulse oximeters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If polysomnography is used for definitive diagnosis of sleep apnea syndrome, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent extracts only the essential vital sign parameters (heart rate, respiratory rate, oxygen saturation, body movement) needed for AHI estimation from the full polysomnography suite, eliminating unnecessary components while maintaining diagnostic capability through machine learning-based analysis of these extracted parameters
Solution Approach 2:
The examination apparatus is designed to perform multiple functions using a single integrated system: it simultaneously monitors heart rate, respiratory rate, oxygen saturation, and body movement, and uses machine learning to estimate AHI, thereby replacing multiple separate devices with one multi-functional unit that reduces overall system complexity
2Measurement precision
If pneumotachograph and pulse oximeter are used for SAS screening, then measurement precision is improved, but ease of operation deteriorates due to subject restraint
Solution Approach 1:
The patent replaces the mechanical pneumotachograph (which requires nasal cannula insertion) with optical detection methods using LEDs and photodetectors to measure respiratory rate through chest/abdomen movement and oxygen saturation through pulse oximetry, thereby eliminating the need for intrusive mechanical components while maintaining measurement accuracy
Solution Approach 2:
The system uses photodetectors to detect optical changes in the subject's tissue caused by blood volume variations during respiration and heartbeat, creating an optical copy of physiological signals that replaces direct mechanical airflow measurement, thus achieving accurate respiratory monitoring without physical contact with the airway
3Device complexity
If simplified examination methods are used for SAS screening, then device complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent introduces machine learning algorithms as an intermediary that processes the collected vital sign data (heart rate, respiratory rate, oxygen saturation, body movement) to estimate AHI, thereby bridging the gap between simplified measurement methods and accurate diagnosis by using computational intelligence to extract meaningful diagnostic information from the collected parameters
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 objective and non-invasive screening of sleep apnea-hypopnea syndrome by accurately estimating AHI from bio-vibration signals without the need for specific medical devices, providing a comprehensive assessment of autonomic nervous activities affected by sleep apnea.
Implementation Method 1
signals obtained with a piezoelectric sensor
Data Source
AI summary
An apnea-hypopnea index is estimated without restraining a subject.Bio-vibration signals are obtained from a subject during sleep in a non-contact and unrestrained manner. From the signals, 4 parameters of respiratory rate, heart rate, body movement, and phase coherence calculated from a difference in instantaneous phase between heartbeat interval variation and respiratory pattern are extracted and put into histograms, which are further transformed into a feature image. The feature image is input in an AHI estimation model that has undergone machine learning.