A Deep Learning Approach to Estimate Multi-Level Mental Stress From EEG Using Serious Games.

Area of Science:

• Neuroscience
• Computer Science
• Psychology

Background:

• Stress impacts task performance and well-being.
• Objective measurement of perceived stress is challenging.
• Existing methods for stress detection have limitations.

Purpose of the Study:

• To evaluate the feasibility of using electroencephalography (EEG) to estimate user-perceived stress during a task.
• To integrate an EEG system with a serious game for stress induction and measurement.
• To apply deep learning (DL) for classifying stress levels based on EEG data.

Main Methods:

• A serious game was developed with increasing difficulty to induce stress.
• An electroencephalography (EEG) system monitored brain activity.
• A recurrent neural network (RNN) with gated recurrent units (GRU) was employed for stress classification.
• The correlation between game difficulty and user stress was assumed.

Main Results:

• The RNN model demonstrated high accuracy in classifying stress levels.
• Accuracy reached up to 94% in certain scenarios, surpassing current state-of-the-art.
• The system successfully correlated game complexity with user stress detection.

Conclusions:

• EEG systems integrated with serious games and DL are effective for predicting mental stress.
• This combined approach offers a promising, accurate method for stress level classification.
• Further research can explore applications in various task-oriented environments.