The Automated Vestibular Rehabilitation System: Normative Data From a Machine Vision-Guided Platform

Vestibular rehabilitation therapy (VRT) is an efficient treatment for dizziness and vertigo, but its accessibility remains limited. This study evaluates the feasibility and performance of a machine vision-based automated alternative-the Automated Vestibular Rehabilitation System (AVRS)-as a more accessible approach to delivering VRT.

Forty age- and sex-matched adults without balance disorders or recent head injury completed one standard VRT exercise, X1, in seated and standing positions using the AVRS, which tracked real-time head and eye movements. Gain, the ratio of eye to head velocity near the neutral head position, was calculated using median values and interquartile ranges. Test-retest reliability (TRTR) was assessed in a subset of 20 participants using intraclass correlation coefficients (ICCs) with 95% confidence intervals.

All participants successfully completed the AVRS-guided vestibular exercises. The n = 20 person test-retest reliability subset returned after 27 ± 9 days. Mean VOR gain approximated the expected physiologic norms (~-1.0) across sessions irrespective of frequency (Session 1: -0.98 ± 0.04; Session 2: -0.99 ± 0.03). Overall reliability between sessions was moderate (ICC (3, 1): 0.59 [95% CI: 0.37-0.75], p < 0.001).

The AVRS demonstrated moderate reliability and accurate VOR gain measurement in healthy adults, supporting its potential as a scalable, accessible tool for VRT delivery, with normative data to guide therapeutic progression in clinical populations.