Automated classification of benign paroxysmal positional vertigo from video-nystagmography using a delay-aware neural network

Nystagmus is a key indicator of vestibular disorders, including benign paroxysmal positional vertigo (BPPV). Accurate diagnosis of BPPV is essential, as it is treatable with specific bedside maneuvers that lead to rapid symptom resolution, thereby improving patient outcomes and reducing unnecessary treatments....

This AI classifier is not yet validated for clinical deployment, but if replicated and externally validated, it could meaningfully reduce misdiagnosis and speed up BPPV canal identification at the bedside without specialist interpretation.

Automated BPPV classification from video-nystagmography could democratize accurate vestibular diagnosis in primary care and emergency settings where specialist expertise is unavailable.

1. 01A delay-aware neural network was trained to classify BPPV type from video-nystagmography (VNG) recordings.
2. 02The model accounts for timing delays in nystagmus signals, a novel architectural feature.
3. 03Automated classification could support diagnosis in settings lacking vestibular specialists.
4. 04Study represents an AI/machine-learning application to a high-prevalence vestibular disorder.
5. 05External validation in diverse clinical populations is needed before clinical adoption.

PMID

42151355

DOI

10.1038/s41598-026-52908-7.

Journal

Scientific Reports

Publication type

research_article

Evidence level

2b

Population

Patients presenting with benign paroxysmal positional vertigo assessed via video-nystagmography

Intervention

Delay-aware neural network for automated BPPV classification from VNG recordings

Comparator

Presumably standard clinical classification or baseline neural network without delay-awareness