JCSE, vol. 20, no. 1, pp.55-62, 2026

DOI: http://dx.doi.org/10.5626/JCSE.2026.20.1.55

A Prediction Method of VR Sickness Susceptibility Using ML

Robin Lee and Yoon Sang Kim
BioComputing Lab., Department of Computer Science and Engineering, Institute for Bioengineering Application Technology, Korea University of Technology and Education, Cheonan, Republic of Korea

Abstract: With the rise of affordable virtual reality (VR) headsets, virtual reality content has become widely accessible, making VR sickness a central concern for user experience. Because symptoms and sensitivity differ among individuals, uniform mitigation strategies can reduce immersion, highlighting the need for personalized approaches. This study proposes a method to predict VR sickness susceptibility using the Motion Sickness Susceptibility Questionnaire (MSSQ) alongside real-time responses measured by the Fast Motion Sickness Scale (FMS). Temporal variations in FMS scores are analyzed to extract features such as slope and frequency, which then serve as inputs to a machine learning model. The model categorizes users’ sensitivity into three levels based on MSSQ percentiles. Experiments involving 50 participants demonstrated the effectiveness of this method, with the Random Forest algorithm achieving the highest prediction accuracy of 80%. By predicting susceptibility prior to VR content exposure, this approach provides a foundation for implementing personalized sickness mitigation strategies that can improve both comfort and immersion in long-form VR experiences.

Keyword: Machine learning; VR sickness; Susceptibility; Prediction

Full Paper:   3 Downloads, 7 View