Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A feature selection method for multilevel mental fatigue EEG classification.

Kai-Quan Shen1, Chong-Jin Ong, Xiao-Ping Li

  • 1Department of Mechanical Engineering, National University of Singapore 117576, Singapore. shen@nus.edu.sg

IEEE Transactions on Bio-Medical Engineering
|July 4, 2007
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Biophysical and Biochemical Assays for Screening Small Molecule Inhibitors Targeting Toxin-Ribosome Interactions.

Toxins·2026
Same author

[Coupling Coordination Measurement and Convergence of Provincial Carbon Emissions and New Quality Productivity in China].

Huan jing ke xue= Huanjing kexue·2026
Same author

Intraductal papillary neoplasm of the biliary tract with typical clinicopathological, endoscopic features: A case report.

World journal of gastrointestinal surgery·2026
Same author

[Mechanism of Xiangshao Granules in alleviating anxiety and depression in mice based on integrated metabolomics and gut microbiota].

Zhongguo Zhong yao za zhi = Zhongguo zhongyao zazhi = China journal of Chinese materia medica·2025
Same author

Two-Dimensional Dual-Switchable Ferroelectric Altermagnets: Altering Electrons and Magnons.

Nano letters·2025
Same author

Role of immunity and inflammation in sarcopenic obesity.

The Journal of nutritional biochemistry·2025
Same journal

Enhancing Volumetric Imaging in Linear-Array Photoacoustic Tomography: multiview fusion with deep learning.

IEEE transactions on bio-medical engineering·2026
Same journal

Robust Rule-based Heuristic Assistance Strategy for a Semi-Active Shoulder Exoskeleton Used in Overhead Work.

IEEE transactions on bio-medical engineering·2026
Same journal

Highly Accelerated 1-mm Isotropic 3D Chemical Exchange Saturation Transfer MRI Using Wave-Co-CAIPI at 5 Tesla.

IEEE transactions on bio-medical engineering·2026
Same journal

Systematic Evaluation of Hip Exoskeleton Assistance Parameters for Enhancing Gait Stability During Ground Slip Perturbations.

IEEE transactions on bio-medical engineering·2026
Same journal

SleepConFormer: A Single-Channel EEG Framework for Sleep Staging and Consciousness Assessment in Patients with Disorders of Consciousness.

IEEE transactions on bio-medical engineering·2026
Same journal

Modeling Partial and Total Support of Left Ventricular Assist Device for Discrete Hemodynamic Control Framework.

IEEE transactions on bio-medical engineering·2026
See all related articles

Recursive feature elimination (RFE) with random forest (RF) improved mental fatigue electroencephalogram (EEG) classification accuracy and reduced features more effectively than initial feature ranking (INIT). RF with RFE achieved a 12.3% error rate using 24 features.

Area of Science:

  • Neuroscience
  • Machine Learning
  • Biomedical Engineering

Background:

  • Mental fatigue significantly impacts cognitive performance and daily activities.
  • Electroencephalogram (EEG) signals offer a non-invasive method for monitoring brain states.
  • Accurate classification of mental fatigue is crucial for applications in driver monitoring, aviation, and demanding work environments.

Purpose of the Study:

  • To compare two feature selection methods, Random Forest with Initial Feature Ranking (RF-INIT) and Random Forest with Recursive Feature Elimination (RF-RFE), for multilevel mental fatigue classification using EEG data.
  • To evaluate the effectiveness of these methods in improving classification performance and reducing the number of features.
  • To identify consistent key features across subjects for robust fatigue detection.

Related Experiment Videos

Main Methods:

  • Utilized electroencephalogram (EEG) time-series data recorded from 12 subjects over 25-hour periods.
  • Applied two feature selection approaches: RF combined with INIT and RF combined with RFE.
  • Employed a "leave-one-proband-out" cross-validation strategy for performance evaluation.
  • Extracted and analyzed features from the EEG data.

Main Results:

  • RF-RFE achieved a lower test error rate (12.3%) using fewer features (24) compared to RF-INIT (15.1% error rate with 64 features).
  • Both methods outperformed using all 304 extracted features (22.1% error rate).
  • RF-RFE identified 17 consistent key features across subjects, demonstrating superior feature reduction and generalizability compared to RF-INIT.

Conclusions:

  • RF-RFE is a more effective approach than RF-INIT for feature selection in mental fatigue EEG classification.
  • The RF-RFE method significantly enhances classification accuracy while substantially reducing the feature set size.
  • The identified consistent features using RF-RFE hold promise for developing reliable and efficient mental fatigue detection systems.