VR-LENS: Super Learning-based Cybersickness Detection and Explainable AI-Guided Deployment in Virtual Reality

Explainability improves trustworthiness in VR health monitoring: The paper demonstrates that using explainable AI methods (SHAP, LIME, and others) to understand why a cybersickness detection system makes its predictions makes the technology more trustworthy and practical for real-world deployment, addressing concerns about "black-box" AI systems.

Simpler models can be as effective as complex ones: By identifying the most important features for detecting cybersickness (eye tracking, player position, and physiological responses like heart rate), researchers reduced model complexity by 1.91X in training time and 2.15X in inference time while maintaining performance—showing that explainability can lead to more efficient AI.

Responsible AI enables practical deployment on consumer devices: The framework successfully deployed cybersickness detection on a standard smartphone processor (Qualcomm Snapdragon 750G), demonstrating that when AI systems are designed with explainability and efficiency in mind from the start, they become accessible and practical for everyday VR users rather than requiring expensive specialized hardware.

Physiological signals are key to understanding VR discomfort: The research identified specific measurable indicators—eye movement, body position, and biological responses—as the dominant factors in detecting cybersickness, providing a scientific foundation for how VR experiences should be designed and monitored for user safety.