• HOME
  • NEWS
  • EXPLORE
    • CAREER
      • Companies
      • Jobs
    • EVENTS
    • iGEM
      • News
      • Team
    • PHOTOS
    • VIDEO
    • WIKI
  • BLOG
  • COMMUNITY
    • FACEBOOK
    • INSTAGRAM
    • TWITTER
Thursday, November 19, 2026
BIOENGINEER.ORG
No Result
View All Result
  • Login
  • HOME
  • NEWS
  • EXPLORE
    • CAREER
      • Companies
      • Jobs
        • Lecturer
        • PhD Studentship
        • Postdoc
        • Research Assistant
    • EVENTS
    • iGEM
      • News
      • Team
    • PHOTOS
    • VIDEO
    • WIKI
  • BLOG
  • COMMUNITY
    • FACEBOOK
    • INSTAGRAM
    • TWITTER
  • HOME
  • NEWS
  • EXPLORE
    • CAREER
      • Companies
      • Jobs
        • Lecturer
        • PhD Studentship
        • Postdoc
        • Research Assistant
    • EVENTS
    • iGEM
      • News
      • Team
    • PHOTOS
    • VIDEO
    • WIKI
  • BLOG
  • COMMUNITY
    • FACEBOOK
    • INSTAGRAM
    • TWITTER
No Result
View All Result
Bioengineer.org
No Result
View All Result
Home NEWS Science News Health

Machine Learning Distinguishes Vestibular Migraine From Posterior Circulation Stroke

by
November 19, 2026
in Health
Reading Time: 6 mins read
0
Machine Learning Distinguishes Vestibular Migraine From Posterior Circulation Stroke

Machine Learning Distinguishes Vestibular Migraine From Posterior Circulation Stroke

Share on FacebookShare on TwitterShare on LinkedinShare on RedditShare on Telegram

A retrospective study of hospitalized patients presenting with acute dizziness or vertigo demonstrated that supervised machine learning models can accurately distinguish vestibular migraine from posterior circulation ischemia using nine routinely accessible clinical variables. Conducted by researchers at The First Affiliated Hospital of Zhengzhou University and published in Frontiers in Neurology, the investigation analyzed clinical histories, videonystagmography metrics, serological panels, and brain magnetic resonance imaging from 571 consecutive patients. The authors addressed a widespread clinical challenge in acute neurology: differentiating benign episodic vestibular conditions from posterior circulation vascular events when both present primarily with vertigo in the absence of obvious focal neurological signs. Across five evaluated algorithms, tree-based models achieved the strongest classification power, with Random Forest and Extreme Gradient Boosting yielding areas under the receiver operating characteristic curve of 0.926 and 0.925, respectively, in an independent testing partition.

Vertigo and dizziness account for a substantial proportion of emergency department presentations, yet diagnostic uncertainty frequently complicates the initial evaluation. Vestibular migraine represents the leading neurological etiology of episodic dizziness in adults, with an estimated population prevalence between 1.0 percent and 2.7 percent, while posterior circulation ischemia constitutes the principal vascular cause of dizzy episodes. Because both disorders often manifest as sudden spinning sensations, nausea, and unsteadiness, misclassification remains frequent. Misidentifying posterior circulation ischemia as vestibular migraine can delay reperfusion therapies and secondary stroke prevention, potentially leading to brainstem infarction, cerebellar compression, or herniation. Conversely, misdiagnosing vestibular migraine as an acute ischemic stroke can expose patients to unnecessary invasive investigations, antithrombotic medications, and bleeding hazards. While bedside protocols such as the three-step ocular motor battery known as HINTS show high diagnostic accuracy for continuous acute vestibular syndrome with nystagmus, their utility is constrained in patients whose symptoms are brief, episodic, or resolved by the time an examination is administered.

To evaluate whether multidimensional clinical profiling could resolve this diagnostic bottleneck, the investigators retrospectively reviewed 4,997 medical records of inpatients admitted with dizziness or vertigo between January 2023 and September 2025. Following the application of strict diagnostic criteria, the analysis focused on 571 patients: 382 diagnosed with vestibular migraine—including 271 with definite and 111 with probable vestibular migraine under international consensus criteria—and 189 diagnosed with posterior circulation ischemia under Bárány Society guidelines. The posterior circulation ischemia cohort was specifically limited to individuals whose initial and predominant presenting manifestation was dizziness or vertigo, excluding patients who arrived with overt neurological signs such as limb weakness, dysarthria, or dysphagia. Ischemic lesions were confirmed via diffusion-weighted magnetic resonance imaging in positive cases, or through demonstration of vertebrobasilar arterial stenosis of at least 50 percent or subsequent objective focal signs during the index hospital stay.

The study recorded broad demographic and vascular risk profiles that highlighted clear baseline differences between the two cohorts. The vestibular migraine group had a mean age of 53.22 years and was 75.4 percent female, whereas patients with posterior circulation ischemia were significantly older, with a mean age of 64.81 years, and were 40.7 percent female. Traditional cardiovascular risk factors were significantly more common among patients with ischemia: hypertension was documented in 75.7 percent versus 49.5 percent of migraine patients, diabetes in 37.6 percent versus 9.9 percent, previous stroke in 31.7 percent versus 4.2 percent, and coronary heart disease in 20.6 percent versus 7.3 percent. Despite these statistical differences, the authors emphasized that nearly half of the vestibular migraine group also suffered from hypertension, demonstrating that the presence of conventional vascular risk factors cannot be used as an exclusive justification for diagnosing posterior circulation stroke in middle-aged or elderly individuals.

Clinical presentations during dizzy episodes also demonstrated marked overlap between the two diagnostic entities. A sensation of rotational visual spinning was reported by 75.1 percent of patients with vestibular migraine and 48.7 percent of patients with posterior circulation ischemia. Accompanying nausea or vomiting was noted in 83.2 percent of migraine cases and 55.6 percent of stroke cases, while positional aggravation occurred in 52.6 percent of migraine cases and 28.0 percent of stroke cases. Although headache accompanying vertigo episodes was significantly more frequent in vestibular migraine at 34.8 percent compared to 22.2 percent in posterior circulation ischemia, its diagnostic utility as an isolated marker was low. The researchers observed that the majority of migraine patients lacked concurrent headaches during vestibular attacks, while headache frequently manifested during acute posterior circulation ischemic events, reinforcing the need for comprehensive quantitative modeling over individual symptom checklists.

Detailed vestibular and audiological assessments identified clear physiological divergence between the cohorts. Objective oculomotor tracking using video nystagmography revealed that central ocular motor abnormalities were substantially more prevalent in posterior circulation ischemia, with abnormal saccades observed in 6.3 percent of ischemia patients compared to 0.8 percent of migraine patients, and abnormal smooth pursuit noted in 9.0 percent versus 0.3 percent. In bithermal caloric testing, vestibular migraine patients exhibited significantly greater horizontal semicircular canal paresis, displaying a median canal paresis value of 37.08 percent compared to 9.38 percent in the ischemia group. Pure-tone audiometry demonstrated that patients with posterior circulation ischemia had worse hearing thresholds across both ears, reflecting older chronological age and cumulative microvascular cochlear damage, despite experiencing fewer subjective complaints of tinnitus or ear fullness than patients with vestibular migraine.

Laboratory blood analyses revealed marked distinctions across glycemic, inflammatory, and metabolic parameters. Fasting blood glucose and glycated hemoglobin levels were significantly elevated in the posterior circulation ischemia group, recording medians of 5.60 millimoles per liter and 6.15 percent, compared to 4.95 millimoles per liter and 5.60 percent in the vestibular migraine group. Plasma homocysteine concentrations were also higher among ischemia cases, with a median of 12.91 micromoles per liter versus 10.21 micromoles per liter in migraine cases. Inflammatory markers derived from complete blood cell counts demonstrated pronounced elevations in ischemia patients: the median neutrophil-to-lymphocyte ratio was 2.52 in the ischemia group compared to 1.68 in vestibular migraine, while the systemic immune-inflammation index was 545.00 versus 379.57, reflecting systemic inflammatory activation associated with vascular atherothrombosis.

Neuroimaging evaluations using brain magnetic resonance imaging incorporated the Fazekas scale to quantify periventricular and deep cerebral white matter hyperintensities on fluid-attenuated inversion recovery sequences. Two independent neurologists graded white matter lesion severity on an aggregate scale ranging from zero to six. Patients with posterior circulation ischemia presented with significantly higher median total Fazekas scores of 3.00, compared to a median score of 2.00 in the vestibular migraine cohort. The authors noted that while chronic migraine has been linked in prior literature to non-specific white matter lesions, the overall extent of structural microvascular parenchymal damage in vestibular migraine remains substantially less severe than that driven by long-standing systemic vascular disease and cerebral ischemia.

To develop a parsimonious diagnostic tool, the total dataset was randomly split using stratified sampling into a training cohort comprising 70 percent of the participants and an independent testing cohort comprising 30 percent, or 172 patients. Within the training set, candidate predictors were refined through an integrated feature selection pipeline combining univariable logistic regression area-under-the-curve filtering, penalized LASSO regression with five-fold cross-validation, and the random-forest-based Boruta algorithm. By taking the intersection of features confirmed across all three algorithmic selection procedures, the researchers isolated nine core variables: the neutrophil-to-lymphocyte ratio, the canal paresis index, total disease duration, the total Fazekas score, patient age, plasma homocysteine, individual episode duration, glycated hemoglobin, and fasting blood glucose.

Five machine learning algorithms were trained on these nine features: Logistic Regression, Support Vector Machine, K-Nearest Neighbors, Random Forest, and Extreme Gradient Boosting. In the test set, all five models demonstrated robust discriminative performance, with areas under the receiver operating characteristic curve spanning from 0.892 for K-Nearest Neighbors to 0.926 for Random Forest. Extreme Gradient Boosting achieved the highest overall diagnostic accuracy at 0.878 and a precision of 0.927, alongside a sensitivity of 0.887 and an F1 score of 0.907. In comparison, Random Forest recorded an accuracy of 0.873, precision of 0.919, sensitivity of 0.887, and an F1 score of 0.903. Decision curve analysis established positive net clinical benefit across decision thresholds ranging from 0.3 to 0.9, while calibration curves demonstrated strong concordance between predicted probabilities and observed outcomes.

To elucidate the decision architecture of the tree-based models, the investigators applied Shapley Additive Explanations to the Extreme Gradient Boosting classifier. Ranking features by mean absolute SHAP values established the neutrophil-to-lymphocyte ratio as the single most influential predictor of classification, followed by canal paresis index, total disease duration, total Fazekas score, and chronological age. Lower neutrophil-to-lymphocyte ratios, higher canal paresis indices, prolonged disease histories, lower Fazekas scores, and younger age drove model predictions toward vestibular migraine. The median disease duration of 365 days in the migraine group versus 30 days in the ischemia group offered strong diagnostic specificity for chronic recurrent disorder, while acute elevations in neutrophil-to-lymphocyte ratio and Fazekas scores shifted predictions strongly toward ischemic vascular etiology.

The authors outlined several methodological limitations that temper immediate clinical translation. Because the study was a retrospective, single-center investigation conducted at a tertiary academic medical center, the enrolled inpatients may exhibit higher illness severity, longer average disease durations, and older age distributions than unselected patients treated in primary care or outpatient clinics. The study design purposefully excluded patients with other common peripheral causes of vertigo, such as benign paroxysmal positional vertigo or Ménière’s disease, meaning the algorithms were trained exclusively on a binary classification task. Furthermore, retrospective documentation lacked granular bedside recordings to evaluate whether spontaneous nystagmus was actively present during initial triage, preventing direct head-to-head performance comparisons with the established HINTS examination protocol. Prospective multicenter studies will be necessary to externally validate these nine-feature algorithms across diverse healthcare settings before widespread clinical implementation.

Subject of Research: Machine learning differentiation of vestibular migraine and posterior circulation ischemia

Article Title: Differentiating vestibular migraine from posterior circulation ischemia initially presenting with dizziness/vertigo: a machine learning approach

Article References: Liu, X., Wang, W., Wang, X., Xiao, X., Wang, Y., Wen, X., & Yang, W. (2026). Differentiating vestibular migraine from posterior circulation ischemia initially presenting with dizziness/vertigo: a machine learning approach. Frontiers in Neurology, 17, Article 1994159. https://doi.org/10.3389/fneur.2026.1994159

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1994159

Keywords: vestibular migraine, posterior circulation ischemia, dizziness, vertigo, machine learning, XGBoost, Random Forest, canal paresis, Fazekas scale, neutrophil to lymphocyte ratio, homocysteine, stroke diagnosis, neuro-otology, white matter hyperintensities, caloric testing, diagnostic models, differential diagnosis

News Source: Ophelia Keating. (November 19, 2026). Machine Learning Distinguishes Vestibular Migraine From Posterior Circulation Stroke. Scienmag.

Tags: caloric testingcanal paresisdiagnostic modelsdifferential diagnosisdizzinessFazekas scalehomocysteineMachine Learningneuro-otologyNeutrophil-to-lymphocyte ratioposterior circulation ischemiaRandom Foreststroke diagnosisvertigovestibular migrainewhite matter hyperintensitiesXGBoost
Share12Tweet7Share2ShareShareShare1

Related Posts

Physical Limitations and Low Social Participation Hinder Older Adults' Exercise Habits

Physical Limitations and Low Social Participation Hinder Older Adults’ Exercise Habits

November 19, 2026
Social Media Dispute Fears Linked to Defensive Emergency Pain Management

Social Media Dispute Fears Linked to Defensive Emergency Pain Management

November 19, 2026

Cortical Plasticity Relies on Distinct Pre- and Postsynaptic NMDA Receptors

November 19, 2026

Analyzing Glycemic Stress Phenotypes in Critically Ill Heart Failure Patients

November 19, 2026

POPULAR NEWS

  • Corncrake Flagship Status Does Not Guarantee Local Recognition in Rural Romania

    Corncrake Flagship Status Does Not Guarantee Local Recognition in Rural Romania

    29 shares
    Share 12 Tweet 7
  • Pre-Rollout Survey Finds Eighteen Percent HPV Vaccine Uptake Among Chinese College Women

    29 shares
    Share 12 Tweet 7
  • Military Physicians Issue Fewer Medical Exemptions as Decision Fatigue Mounts Across Workdays

    29 shares
    Share 12 Tweet 7
  • Counseling and Mobile App Increase Postpartum Modern Contraceptive Uptake in Jordan

    29 shares
    Share 12 Tweet 7

About

We bring you the latest biotechnology news from best research centers and universities around the world. Check our website.

Follow us

Recent News

Physical Limitations and Low Social Participation Hinder Older Adults’ Exercise Habits

Dairy Processing Study in India Evaluates Milk Quality and Microbial Safety

Social Media Dispute Fears Linked to Defensive Emergency Pain Management

Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 88 other subscribers
  • Contact Us

Bioengineer.org © Copyright 2023 All Rights Reserved.

Welcome Back!

Login to your account below

Forgotten Password?

Retrieve your password

Please enter your username or email address to reset your password.

Log In
No Result
View All Result
  • Homepages
    • Home Page 1
    • Home Page 2
  • News
  • National
  • Business
  • Health
  • Lifestyle
  • Science

Bioengineer.org © Copyright 2023 All Rights Reserved.