Reliability Testing of a Multitask Virtual Reality Sensorimotor Assessment.

Abstract

We evaluated the test-retest reliability of a comprehensive, multitask virtual reality (VR)-based sensorimotor assessment designed to capture and quantify aspects of body movement during standardized visuomotor, auditory, gait, oculomotor, and balance tasks. Twenty healthy adults completed 3 sessions of the Dynamic Assessment of Sensorimotor Health protocol on nonconsecutive days at a university research center. Dynamic Assessment of Sensorimotor Health includes 7 VR tasks (Trail Making, Sound Localization, Tandem Gait, Reaction Board, Smooth Pursuit, Single-Leg Balance, and Target Matching Balance). Session-level performance metrics were derived from head-, hand-, and eye-tracking data, and intraclass correlation coefficients (ICC[3,1]) and session means were calculated for each metric. Reliability ranged from poor to excellent (ICC = .07 to .81). Metrics from the Reaction Board (reaction time) and Target Matching Balance (head path length) tasks demonstrated the highest reliability (ICC up to .81), and several measures from the Trail Making, Smooth Pursuit, and Sound Localization tasks showed fair-to-good reliability. These findings identify specific VR-derived body movement metrics that show promise for tracking sensorimotor performance over time and highlight others that may require redesign or alternative processing before use in longitudinal or clinical applications.