Related Experiment Video
Updated: Aug 13, 2025
06:51
PIPEMAT-RS: Development and Validation of a Standardized MATLAB Pipeline for Resting-State EEG Preprocessing
Published on: June 6, 2025
290
Assessment of Model Accuracy in Eyes Open and Closed EEG Data: Effect of Data Pre-Processing and Validation Methods
Jamolbek Mattiev1, Jakob Sajovic2,3, Gorazd Drevenšek3,4
1Department of Information Technologies, Urgench State University, Khamid Alimdjan 14, Urgench 220100, Uzbekistan.
Bioengineering (Basel, Switzerland)
|January 21, 2023
Summary
Simple preprocessing improves electroencephalography (EEG) classification for individual subjects, while comprehensive preprocessing aids subject-invariant classification. Choosing the right method depends on the intended application.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Machine Learning
Background:
- Electroencephalography (EEG) data is crucial for validating human brain activity classification methods.
- Current validation often uses cross-validation on minimally preprocessed EEG data, which can be affected by noise and overfitting.
- Limited dataset sizes often necessitate dividing single-subject EEG data, potentially impacting model generalizability.
Purpose of the Study:
- To investigate the impact of preprocessing techniques (simple vs. comprehensive) on classification accuracy using EEG data.
- To compare the effectiveness of cross-validation and hold-out validation methods for EEG data.
- To assess the influence of subject-specific versus subject-invariant classification goals on preprocessing choices.
Main Methods:
- Tested fourteen classification algorithms from WEKA and MATLAB on EEG data from 50 subjects (eyes open/closed).
- Compared classification accuracy using both simple and comprehensive preprocessing methods.
- Employed both cross-validation and hold-out validation strategies to evaluate model performance.
Main Results:
- Simple preprocessing yielded superior results in cross-validation testing, indicating better subject-specific performance.
- Comprehensive preprocessing showed better performance in hold-out validation, suggesting advantages for subject-invariant classification.
- Mixing subject data in training and testing sets significantly increased hold-out accuracy, highlighting overfitting issues.
Conclusions:
- The choice between simple and comprehensive EEG data preprocessing depends on whether the goal is subject-specific or subject-invariant classification.
- Comprehensive preprocessing is beneficial for achieving subject-invariant classification accuracy.
- Simple preprocessing can lead to higher subject-specific classification accuracy, but researchers must be aware of potential overfitting.
