Preliminary Evaluation of the Virtual Reality-Based Gait Sensory Interaction Test (GaitSIT) for Quantifying Sensory...

Background: Walking is a dynamic activity that relies on inputs from the multisensory system, i.e., somatosensory, vision, and vestibular. These inputs are processed and integrated in the central nervous system to produce motor impulses for efficient walking balance. The Sensory Organization Test (SOT) is established as the gold standard for assessing sensory contributions to standing balance. However, no comparable assessments have been developed for the clinical evaluation of balance during gait. This study evaluated the Gait Sensory Interaction Test (GaitSIT), a novel virtual reality (VR)-based assessment for characterizing sensory-condition-specific changes in walking balance. Methods: The GaitSIT comprises a VR environment with a physical compliant foam walking surface that evaluates gait-balance by systematically manipulating and evaluating the sensory systems. Twenty-nine healthy young adults (mean age 24.9 ± 6.4 years) were instructed to complete 6 m walking trials under six standardized conditions (C): eyes open, eyes closed/dark scene, and rotating visual scenes on a firm surface, then repeated on a foam surface. Wearing an Oculus VR headset, participants were instructed to walk in a straight line at their preferred speed, as naturally as possible, in two test sessions on the same day, followed by a third test session 24 h later. Headset-derived sway measures, including position, velocity, and acceleration data, were recorded, and the continuous trajectory deviation angle (i.e., directional control) and sensory ratios were calculated. Linear mixed-effects models included trial-level walking speed as a covariate. Additionally, participants completed the modified Clinical Test of Sensory Interaction on Balance (mCTSIB) as a clinical standing-balance reference measure; its concurrent-validity findings will be reported separately. Results: Significant condition effects were observed for position, velocity, acceleration, and CTDA after adjustment for trial-level walking speed (all p<0.001), indicating that the six sensory conditions elicited distinct gait-balance responses. Significant differences relative to the baseline condition (C1) were observed across conditions C2-C6 for position, C3-C6 for velocity, and C2 and C5 for acceleration. Session effects were not significant for any primary kinematic outcome after speed adjustment. A significant condition × session interaction was observed for position (p<0.001), whereas velocity, acceleration, and CTDA demonstrated no significant interactions. Walking speed was significantly associated with position, acceleration, and CTDA, but not velocity. Sensory-ratio analyses revealed larger visual and vestibular ratios relative to somatosensory ratios, with the visual and vestibular ratios generally decreasing across sessions. Conclusions: GaitSIT successfully manipulated sensory conditions during overground walking and produced significant changes in gait-related sway, directional control, and sensory-ratio measures. These findings support the feasibility of GaitSIT as a portable, low-cost, and immersive assessment framework for characterizing sensory-condition-specific gait-balance responses after accounting for walking speed and providing indirect behavioral indices related to sensory reweighting.