Acrophobia Quantified by EEG Based on CNN Incorporating Granger Causality

Area of Science:

• Neuroscience
• Psychology
• Virtual Reality

Background:

• Acrophobia (fear of heights) is difficult to quantify with traditional methods.
• Accurate assessment is crucial for high-altitude worker safety.
• Existing methods lack precision in evaluating acrophobia levels.

Purpose of the Study:

• To develop a reliable method for quantifying acrophobia.
• To provide safety recommendations for individuals working at high altitudes.
• To differentiate between no acrophobia, moderate acrophobia, and severe acrophobia.

Main Methods:

• Simulated a virtual reality environment: High-altitude Plank Walking Challenge.
• Proposed the Granger Causality Convolutional Neural Network (GCCNN) method.
• Analyzed noninvasive scalp electroencephalogram (EEG) signals using GCCNN for classification.

Main Results:

• GCCNN achieved 98.74% accuracy in a two-class task (acrophobia vs. no acrophobia).
• GCCNN achieved 98.47% accuracy in a three-class task (no, moderate, severe acrophobia).
• Demonstrated superior classification performance compared to mainstream methods.

Conclusions:

• The proposed GCCNN method offers a highly accurate quantitative assessment of acrophobia.
• This method has strong potential for practical applications in occupational safety.
• Enhanced quantification of acrophobia can inform targeted safety interventions for high-altitude workers.