A Generalized Model to Estimate Reaction Time Corresponding to Visual Stimulus Using Single-Trial EEG.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Estimating reaction time (RT) from brain signals is challenging due to trial-to-trial variability in perceptual decision-making.
• Existing electroencephalogram (EEG) studies for RT estimation are often subject-specific and rely solely on regression.
• Accurate RT estimation is crucial for developing effective brain-computer interfaces (BCIs).

Purpose of the Study:

• To introduce a generalized, single-trial EEG-based model for estimating reaction time (RT) in a simple visual reaction task.
• To explore both regression and classification approaches for RT estimation using EEG features.
• To identify relevant EEG channels and features for improved RT prediction.

Main Methods:

• A generalized model was developed to estimate RT using single-trial EEG features.
• Both regression and classification (binary and 3-class) models were implemented.
• EEG channel relevance and feature importance were analyzed for RT estimation.

Main Results:

• The regression model predicted RT with a root mean square error of 111.2 ms and a correlation of 0.74.
• Classification models achieved 79% (binary) and 72% (3-class) accuracy.
• The model achieved 95% accuracy when classifying only high and low RT groups.

Conclusions:

• A generalized EEG-based model can effectively estimate reaction time (RT) using single-trial data.
• Both regression and classification approaches show promise for RT estimation from EEG.
• This research contributes to advancing EEG signal analysis for BCIs, particularly for individuals with neuromuscular disorders.