Related Experiment Videos
Automated detection and classification of sleep-disordered breathing from conventional polysomnography data
B H Taha1, J A Dempsey, S M Weber
1Department of Preventive Medicine, University of Wisconsin-Madison 53705, USA.
Sleep
|February 11, 1998
Summary
This study developed an efficient algorithm for detecting sleep-disordered breathing (SDB) using pulse oximetry and polysomnography (PSG) data. The algorithm accurately identifies apneas and hypopneas, improving automated SDB detection.
Area of Science:
- Sleep Medicine
- Respiratory Physiology
- Biomedical Engineering
Background:
- Automated detection of sleep-disordered breathing (SDB) from polysomnography (PSG) is challenging due to indirect breathing measurements.
- Integrating pulse oximetry data can enhance SDB event detection algorithms.
Purpose of the Study:
- To develop and evaluate an efficient algorithm for automated detection and classification of SDB events.
- To incorporate pulse oximetry into SDB definitions to overcome limitations of indirect breathing measurements.
Main Methods:
- Developed an algorithm using respiratory inductive plethysmography (RIP) and pulse oximetry to define and detect apneas and hypopneas.
- Apnea detection: RIP cessation (≥10s) coincident with desaturation (≥2% fall).
- Hypopnea detection: ≥3 breaths with ≥20% RIP reduction, followed by return to ≥90% baseline, during desaturation.
Main Results:
- The algorithm demonstrated strong event-by-event agreement with manual scoring (1,938 SDB events across 10 PSG records).
- Sensitivity and specificity for apneas were 73.6% and 90.8%; for hypopneas, 84.1% and 86.1%.
- Overall, the algorithm detected 93.1% of manually identified SDB events.
Conclusions:
- An efficient algorithm for detecting and classifying SDB events has been designed.
- The algorithm effectively emulates manual scoring with high accuracy, integrating pulse oximetry for improved detection.
- This approach enhances automated SDB analysis from PSG data.