Print Close

Machine-Learning-Based Classification of Patients with Obstructive Sleep Apnea using Sleep EEG

Poster No:

2098

Submission Type:

Abstract Submission

Authors:

Ji Yong Park¹, Dongyeop Kim², Young Wook Song¹, Euijin Kim³, Eun Yeon Joo⁴, Sungkean Kim¹^,3

Institutions:

¹Department of Applied Artificial Intelligence, Hanyang University, Ansan, Korea, Republic of, ²Department of Neurology, Seoul Hospital, Ewha Womans University College of Medicine, Seoul, Korea, Republic of, ³Department of Human-Computer Interaction, Hanyang University, Ansan, Korea, Republic of, ⁴Department of Neurology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea, Republic of

First Author:

Ji Yong Park
Department of Applied Artificial Intelligence, Hanyang University
Ansan, Korea, Republic of

Co-Author(s):

Dongyeop Kim
Department of Neurology, Seoul Hospital, Ewha Womans University College of Medicine
Seoul, Korea, Republic of

Young Wook Song
Department of Applied Artificial Intelligence, Hanyang University
Ansan, Korea, Republic of

Euijin Kim
Department of Human-Computer Interaction, Hanyang University
Ansan, Korea, Republic of

Eun Yeon Joo
Department of Neurology, Samsung Medical Center, Sungkyunkwan University School of Medicine
Seoul, Korea, Republic of

Sungkean Kim
Department of Applied Artificial Intelligence, Hanyang University|Department of Human-Computer Interaction, Hanyang University
Ansan, Korea, Republic of|Ansan, Korea, Republic of

Introduction:

Obstructive sleep apnea (OSA) is a widespread sleep disorder characterized by oxygen desaturation, frequent arousal, and increased sympathetic activity caused by continuous upper airway obstruction. Electroencephalography (EEG) obtained from the polysomnography (PSG) system is mainly used for scoring sleep stages and arousals induced by various events, including apnea or hypopnea. However, there were some studies on classifying OSA patients and healthy controls by analyzing the sleep EEG from PSG with six channels. This study aimed to investigate the possibility of sleep EEG with 19 channels as a biomarker of OSA by discriminating between the normal to mild OSA group and the moderate to severe OSA group using machine learning techniques.

Methods:

Overnight PSG test was conducted on 18 participants with normal to mild OSA (AHI<15) and 20 participants with moderate to severe OSA (AHI≥15) using an Embla N7000 (Medcare-Embla®, Reykjavik, Iceland). EEG data were recorded with 19 electrodes in accordance with the international 10-20 system. Relative power from power spectral analysis, graph theory-based weighted network indices including strength, clustering coefficient, path length, and eigenvector centrality (EC), and microstate parameters including global explained variance, duration, occurrence, coverage, and transition probabilities were compared between two groups (Brodbeck, 2012). In addition, correlation analyses between EEG features and polysomnographic parameters were performed. Furthermore, we applied machine learning techniques to discriminate between the two groups using EEG features. We used the sequential forward selection method to find the best feature subset with the AdaBoost classifier. Ten-fold cross-validation was repeated ten times to estimate accuracy, sensitivity, and specificity.

Results:

There were no significant differences in age, sex, and BMI between the two groups. The feature set was constituted of 71 EEG features including relative power (n = 48), network indices (n = 21), and microstate parameters (n = 2) showing significant differences between the two groups. When the best classification performances were detected, the result showed an 88.3% accuracy, 92.0% sensitivity, and 84.0% specificity with seven features including the theta power at C4, P3, and P4 electrodes during NREM1, the gamma power at F8 electrode during NREM1 and Pz electrode during REM, the gamma EC at F7 electrode during REM, and duration of microstate 4 during NREM2. Mainly, the theta activities during NREM1 were decreased, and the gamma activities during NREM1 and REM were increased in the moderate to severe OSA group. The selected features showed significant correlations with various polysomnographic parameters.

·Microstate topographies of the total participants. MS, microstate.

Conclusions:

In this study, we investigated the differences in EEG measures during sleep between the normal to mild OSA and moderate to severe OSA groups. Also, we classified the two groups with reasonable classification performances based on the EEG features. Our results suggest that EEG measures might serve clinically as biomarkers of OSA.

Modeling and Analysis Methods:

EEG/MEG Modeling and Analysis ²

Perception, Attention and Motor Behavior:

Sleep and Wakefulness ¹

Keywords:

Electroencephaolography (EEG)

Machine Learning

Sleep

Other - Obstructive sleep apnea;Microstate

^1|2Indicates the priority used for review

Abstract Information

By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.

I accept

The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:

I do not want to participate in the reproducibility challenge.

Please indicate below if your study was a "resting state" or "task-activation” study.

Other

Healthy subjects only or patients (note that patient studies may also involve healthy subjects):

Patients

Was this research conducted in the United States?

Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.

Yes

Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.

Not applicable

Please indicate which methods were used in your research:

EEG/ERP

Provide references using APA citation style.

Brodbeck, V., Kuhn, A., von Wegner, F., Morzelewski, A., Tagliazucchi, E., Borisov, S., ... & Laufs, H. (2012). EEG microstates of wakefulness and NREM sleep. Neuroimage, 62(3), 2129-2139.

UNESCO Institute of Statistics and World Bank Waiver Form

I attest that I currently live, work, or study in a country on the UNESCO Institute of Statistics and World Bank List of Low and Middle Income Countries list provided.