1361-EX Stress Onset Prediction in Virtual Environments Using Physiological Data

Location

https://campus.kennesaw.edu/colleges-departments/ccse/events/computing-showcase/fa26-cday-program.php

Document Type

Event

Start Date

20-11-2026 2:00 PM

Description

Mental stress hampers the task performance significantly in virtual environments. This study investigated whether machine learning could predict the onset of stress-related physiological responses in virtual reality using electrodermal activity (EDA) and muscle activity data. A Random Forest model was trained and evaluated using the publicly available Virtual Reality Balance Disturbance Dataset, which included 12 participants. Eighteen EDA increase thresholds were tested, along with different combinations of eight muscle groups. The 0.04 μS threshold achieved the highest mean accuracy (54.22%) and F1-score (56.67%), while 0.065 μS produced the highest balanced accuracy (54.81%). Muscle-group differences were not statistically significant. Overall, the findings demonstrate limited predictive performance but provide insight into using physiological and movement data to identify the onset of stress-related responses.

This document is currently not available here.

Share

COinS
 
Nov 20th, 2:00 PM

1361-EX Stress Onset Prediction in Virtual Environments Using Physiological Data

https://campus.kennesaw.edu/colleges-departments/ccse/events/computing-showcase/fa26-cday-program.php

Mental stress hampers the task performance significantly in virtual environments. This study investigated whether machine learning could predict the onset of stress-related physiological responses in virtual reality using electrodermal activity (EDA) and muscle activity data. A Random Forest model was trained and evaluated using the publicly available Virtual Reality Balance Disturbance Dataset, which included 12 participants. Eighteen EDA increase thresholds were tested, along with different combinations of eight muscle groups. The 0.04 μS threshold achieved the highest mean accuracy (54.22%) and F1-score (56.67%), while 0.065 μS produced the highest balanced accuracy (54.81%). Muscle-group differences were not statistically significant. Overall, the findings demonstrate limited predictive performance but provide insight into using physiological and movement data to identify the onset of stress-related responses.