JMIR Publications

ABSTRACT

Background:

Accurate nystagmus examination is essential for distinguishing potentially life-threatening central causes of vertigo (e.g., stroke) from benign peripheral vestibular dysfunction. Conventional video-oculography (VOG) systems—while sensitive and quantitative—require dedicated laboratories, bulky equipment, and specialized operators, limiting rapid triage in outpatient clinics, emergency departments, and telemedicine. Wearable augmented reality (AR)–based platforms may overcome these barriers by providing portable, automated, and user-friendly real-time oculomotor assessment with secure cloud-based data transmission.

Objective:

To evaluate the preliminary diagnostic comparability of a wearable AR–based nystagmus examination system with conventional VOG assessment in detecting oculomotor abnormalities, including suspected central vestibular pathology, among adult vertigo patients.

Methods:

This evaluator‐blinded, randomized crossover feasibility study (IRB No. 202202194B0C501) was conducted from October 2024 to January 2025 at Kaohsiung Chang Gung Memorial Hospital, Taiwan. The AR system comprised J7EF Gaze smart glasses tethered to an Android‐based portable device running in‐house Unity 3D software, delivering six standardized oculomotor stimuli—horizontal/vertical gaze‐evoked fixation, saccades, and smooth pursuit—mirroring the VOG protocol. All eye movement data were encrypted and transmitted to a secure cloud platform for automated processing and retrospective review. Nine adult outpatients with clinically diagnosed vertigo were recruited; eight completed paired AR‐ and VOG‐based assessments, separated by a 30-minute washout. All 48 paired waveform outputs were pooled, randomized, and blindly interpreted by a single clinician. The primary outcome—diagnostic concordance—was quantified as percent agreement. Secondary outcomes included patient discomfort (Visual Analog Scale, 0–10), examination duration, and diagnostic performance metrics (accuracy, sensitivity, specificity, positive predictive value [PPV], negative predictive value [NPV]) with 95% confidence intervals (Clopper–Pearson exact method).

Results:

Eight participants (mean age 60.4 years; range 46–72) completed the study with no significant adverse events or discomfort. One participant with a history of cataract surgery could not be calibrated by the AR glasses, resulting in an unsuccessful AR examination. Agreement rates across signal dimensions ranged from 62.5% to 87.5%, yielding an overall accuracy of 77.1%. Sensitivity was 81.8% (9/11; 95% CI 48.2–97.7%), specificity 75.7% (28/37; 95% CI 58.8–88.2%), PPV 50.0% (9/18; 95% CI 26.0–73.9%), and NPV 93.3% (28/30; 95% CI 77.9–99.2%). For central pathology detection, sensitivity reached 83.3% and specificity 100%. One participant with prior cataract surgery could not be calibrated, highlighting the need for adaptive algorithms.

Conclusions:

The wearable AR–based nystagmus examination system demonstrated preliminary diagnostic comparability to VOG, high NPV, reduced examination time, and strong tolerability, supporting its potential as a portable screening and triage tool for central vestibular disorders in diverse clinical and remote settings. Moderate PPV, calibration challenges, and wide confidence intervals underscore the need for algorithm refinement, larger multicenter validation, and integration with AI-driven decision support.