Related Experiment Video
Updated: Aug 5, 2025
06:37
Electroencephalography Network Indices as Biomarkers of Upper Limb Impairment in Chronic Stroke
Published on: July 14, 2023
939
Prediction of Cognitive Load from Electroencephalography Signals Using Long Short-Term Memory Network
Gilsang Yoo1, Hyeoncheol Kim2, Sungdae Hong3
1Creative Informatics and Computing Institute, Korea University, Seoul 02841, Republic of Korea.
Bioengineering (Basel, Switzerland)
|March 29, 2023
Summary
This study introduces a deep learning method using electroencephalography (EEG) signals to detect confusion, achieving 87.1% accuracy. This advance supports personalized adaptive learning systems by measuring cognitive load in real time.
Area of Science:
- Neuroscience
- Machine Learning
- Educational Technology
Background:
- Adaptive learning models require real-time measurement of cognitive load.
- Confusion, or brain fog, negatively impacts performance and is difficult to detect.
- Applications include online education and driver fatigue monitoring.
Purpose of the Study:
- To develop a deep learning method for recognizing cognitive load using electroencephalography (EEG) signals.
- To evaluate the proposed method against traditional machine learning algorithms.
- To enable real-time detection of confusion for adaptive systems.
Main Methods:
- Utilized a long short-term memory (LSTM) network with an attention mechanism for EEG signal analysis.
- Collected EEG data from a brainwave information database with associated mental load data.
- Compared LSTM performance against Random Forest, AdaBoost, SVM, XGBoost, and ANN models.
Main Results:
- The proposed LSTM model achieved the highest accuracy at 87.1%.
- Comparative algorithms showed lower accuracies: Random Forest (64%), AdaBoost (64.31%), SVM (60.9%), XGBoost (67.3%), and ANN (71.4%).
- The deep learning approach significantly outperformed traditional methods.
Conclusions:
- The developed deep learning method effectively recognizes cognitive load from EEG signals.
- This research paves the way for personalized adaptive learning systems.
- Real-time cognitive load measurement using portable EEG systems is feasible.
