Separation of stroke from vestibular neuritis using the video head impulse test: machine learning models versus expert clinicians

Aug 4, 2026

Upload by Skyler

Acute vestibular syndrome can reflect either benign vestibular neuritis (VN) or posterior circulation stroke (PCS), which requires urgent recognition and treatment. The video head impulse test (VHIT) quantitatively measures the vestibulo-ocular reflex and can distinguish between these diagnoses, but routine use depends on expert interpretation. Machine learning models were developed to differentiate PCS from VN using only unedited acute VHIT head-and-eye velocity data, then validated on an independent dataset and compared with expert clinicians and a standard gain cutoff metric.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Abstract

Methods

Results

Conclusion

Introduction

Journal of Neurology (2025) 272:248 [https://doi.org/10.1007/s00415-025-12918-3](https://doi.org/10.1007/s00415-025-12918-3) Separation of stroke from vestibular neuritis using the video head impulse test: machine learning models versus expert clinicians Chao Wang1,2 · Jeevan Sreerama1 · Benjamin Nham3 · Nicole Reid2 · Nese Ozalp4 · James O. Thomas4 · Cecilia Cappelen‑Smith4,5 · Zeljka Calic4,5 · Andrew P. Bradshaw2 · Sally M. Rosengren1,2 · Gülden Akdal7,8 · G. Michael Halmagyi1,2 · Deborah A. Black6 · David Burke1,2 · Mukesh Prasad9 · Gnana K. Bharathy9 · Miriam S. Welgampola1,2 Received: 10 November 2024 / Revised: 9 January 2025 / Accepted: 12 January 2025 / Published online: 5 March 2025 © The Author(s) 2025 Abstract Background Acute vestibular syndrome usually represents either vestibular neuritis (VN), an innocuous viral illness, or posterior circulation stroke (PCS), a potentially life-threatening event. The video head impulse test (VHIT) is a quantitative measure of the vestibulo-ocular reflex that can distinguish between these two diagnoses. It can be rapidly performed at the bedside by any trained healthcare professional but requires interpretation by an expert clinician. We developed machine learning models to differentiate between PCS and VN using only the VHIT. Methods We trained machine learning classification models using unedited head-and eye-velocity data from acute VHIT performed in an Emergency Room on patients presenting with acute vestibular syndrome and whose final diagnosis was VN or PCS. The models were validated using an independent test dataset collected at a second institution. We compared the performance of the models against expert clinicians as well as a widely used VHIT metric: the gain cutoff value. Results The training and test datasets comprised 252 and 49 patients, respectively. In the test dataset, the best machine learning model identified VN with 87.8%(95% CI 77.6%–95.9%) accuracy. Model performance was not significantly different (p = 0.56) from that of blinded expert clinicians who achieved 85.7% accuracy (75.5%–93.9%) and was superior (p = 0.01) to that of the optimal gain cutoff value (75.5% accuracy (63.8%–85.7%)) . Conclusion Machine learning models can effectively differentiate PCS from VN using only VHIT data, with comparable accuracy to expert clinicians. They hold promise as a tool to assist Emergency Room clinicians evaluating patients with acute vestibular syndrome. Keywords Stroke · Vestibular neuritis · Artificial intelligence · Machine learning · Video head impulse test * Miriam S. Welgampola [miriam@icn.usyd.edu.au](miriam@icn.usyd.edu.au) 1 Central Clinical School, University of Sydney, Sydney, NSW, Australia 2 Institute of Clinical Neurosciences, Royal Prince Alfred Hospital, Sydney, NSW, Australia 3 St George and Sutherland Clinical School, University of New South Wales, Sydney, NSW, Australia 4 Department of Neurophysiology, Liverpool Hospital, Sydney, NSW, Australia 5 South Western Sydney Clinical School, University of New South Wales, Sydney, NSW, Australia 6 Faculty of Medicine and Health, University of Sydney, Sydney, NSW, Australia 7 Department of Neurosciences, Institute of Health Sciences, Dokuz Eylül University, Izmir, Türkiye 8 Department of Neurology, Faculty of Medicine, Dokuz Eylül University, Izmir, Türkiye 9 School of Computer Science, Faculty of Engineering and Information Technology, University of Technology Sydney, Sydney, NSW, Australia Introduction The acute vestibular syndrome (AVS) refers to sudden onset severe, persistent vertigo and/or imbalance [1] and is a common presentation to the Emergency Room. Correctly identifying the cause is important as it is usually due to one of two conditions with very different therapeutic and prognostic implications: posterior circulation stroke (PCS), which is potentially life-threatening and may necessitate urgent reperfusion therapy, or vestibular neuritis (VN),

FAQs

What problem does this study address in acute vestibular syndrome?

It targets the challenge of distinguishing posterior circulation stroke (PCS) from vestibular neuritis (VN), two conditions with very different urgency, treatment, and outcomes.

How do the machine learning models use the video head impulse test?

They are trained to classify PCS versus VN using only VHIT data, specifically unedited head-and-eye velocity signals from acute VHIT recordings.

How does model performance compare with expert clinicians and the gain cutoff metric?

In the test dataset, the best model identified VN with 87.8% accuracy, which was not significantly different from blinded expert clinicians (85.7%) and was superior to the optimal gain cutoff value (75.5%).

Aug 4, 2026

Upload by Skyler