1 January 2026 in Journal on Interactive Systems
Cybersickness (CS) is one of the main challenges for the adoption of Virtual Reality (VR), manifesting through symptoms such as nausea, dizziness, and eye strain, particularly in Head-Mounted Display (HMD) devices. Although subjective measures, such as questionnaires, are widely used to assess CS, they do not allow for real-time user feedback. This study investigates the role of biosignals in identifying the causes and indicators of CS in VR games, employing Symbolic Machine Learning (SML) to classify the most relevant factors. Our approach combines Electrocardiogram (ECG), Electrodermal Activity (EDA), and Accelerometer (ACC) data with game metrics and user profile attributes. Data were collected from two VR games: a car game and a flying game. Decision Trees and Random Forests were used to build interpretable models, and the results showed that integrating biosignals and game data significantly improves CS prediction, with Random Forest achieving an AUC of 0.95. The findings highlight that exposure time, motion intensity, and electrodermal activity are among the key predictors of CS, reinforcing the importance of physiological monitoring in VR research. Furthermore, the study demonstrates the potential of SML in creating explainable models, contributing to more effective strategies for mitigating CS.