A Machine Learning Approach to Classifying EEG Data Collected with or without Haptic Feedback during a Simulated...

Area of Science:

• Neuroscience
• Computer Science
• Biomedical Engineering

Background:

• Virtual reality (VR) and AI simulations offer accessible training platforms.
• Objective neural assessment using electroencephalography (EEG) is preferred over self-reports for sensory feedback analysis.
• Identifying specific EEG features related to sensory feedback impact remains a challenge.

Purpose of the Study:

• To apply machine learning to differentiate neural circuitry associated with haptic versus non-haptic feedback during a simulated drilling task.
• To identify key EEG signal features indicative of haptic feedback's influence on neural processing.

Main Methods:

• Machine learning techniques were employed to analyze EEG data from a simulated drilling task.
• Nine EEG channels were analyzed, extracting 360 features (time-domain, frequency-domain, nonlinear).
• A feature selection process identified the most relevant EEG features for differentiating feedback types.

Main Results:

• Machine learning models accurately distinguished trials with haptic feedback from those without, achieving over 90% accuracy with five selected features.
• Accuracy increased to 99% when utilizing ten features.
• Key identified features included Hurst exponent, kurtosis, power spectral density, variance, and spectral entropy across specific frequency bands.

Conclusions:

• Machine learning effectively predicts the impact of haptic feedback on neural processing during VR simulation.
• This approach can optimize VR and simulation-based training for skill acquisition.
• Future research can leverage these findings to refine VR training protocols and objective assessments.