Advanced Sensing System for Sleep Bruxism across Multiple Postures via EMG and Machine Learning
Jahan Zeb Gul1, Noor Fatima2, Zia Mohy Ud Din2
1Department of Electronic Engineering, Maynooth University, W23A3HY Maynooth, Ireland.
Sensors (Basel, Switzerland)
|August 29, 2024
Summary
This study introduces a novel approach to diagnosing sleep bruxism using electromyography (EMG) data from the temporalis muscle. Machine learning models achieved high accuracy, particularly in the left lateral position, improving bruxism detection.
Area of Science:
- Neurology
- Biomedical Engineering
- Data Science
Background:
- Diagnosing sleep bruxism is challenging due to the difficulty in distinguishing true bruxism from normal muscle activity.
- Current detection methods, including electromyography (EMG), electrocardiography (ECG), and electroencephalography (EEG), have limitations in accessibility and effectiveness.
- Previous research primarily focused on the masseter muscle and supine sleep positions, potentially overlooking other key indicators.
Purpose of the Study:
- To investigate the efficacy of electromyography (EMG) in detecting sleep bruxism by analyzing the temporalis muscle activity.
- To evaluate machine learning (ML) classifiers for identifying sleep bruxism across different sleep positions (supine, left lateral, right lateral).
- To enhance the accuracy and clinical applicability of sleep bruxism diagnosis.
Main Methods:
- Collected EMG data measuring maximum voluntary contraction of temporalis and masseter muscles in supine, left lateral, and right lateral positions.
- Extracted 10 time-domain features from the EMG signals.
- Compared the performance of six machine learning classifiers, including Random Forest, for sleep bruxism detection.
Main Results:
- Machine learning models demonstrated superior accuracy in detecting sleep bruxism using temporalis muscle data compared to masseter muscle data.
- The Random Forest classifier achieved the highest performance among the evaluated models.
- An accuracy of 93.33% was specifically achieved for sleep bruxism detection in the left lateral recumbent position.
Conclusions:
- The temporalis muscle is a promising target for improving sleep bruxism detection accuracy.
- Machine learning, particularly Random Forest, shows significant potential for enhancing the clinical diagnosis of sleep bruxism.
- This research offers a pathway towards more effective, data-driven management of sleep bruxism.
