A multimodal data-based diagnostic model for predicting vestibular migraine: a retrospective study

Area of Science:

• Neurology
• Ophthalmology
• Audiology

Background:

• Vestibular migraine (VM) presents with vertigo and migraine symptoms, often leading to misdiagnosis due to its varied presentation and lack of biomarkers.
• Accurate diagnosis of VM is challenging, impacting patient management and treatment strategies.

Purpose of the Study:

• To develop and validate a diagnostic prediction model for VM using multimodal data.
• To enhance diagnostic accuracy for vestibular migraine.

Main Methods:

• A cohort of 288 patients (141 VM, 147 controls) was analyzed.
• Multimodal data included clinical features, vestibular tests, blood work, echocardiography, and psychological assessments.
• Logistic regression and ROC curve analysis were employed to build and evaluate the model.

Main Results:

• VM patients differed from controls in demographics, symptoms (photophobia, phonophobia, tinnitus, emotional triggers, insomnia), and medical history.
• Distinct patterns in vestibular function, hematological indicators (lower vitamin D, D-dimer; higher platelets, calcium), and prevalence of right-to-left shunt were observed in VM patients.
• A six-variable model (BMI, emotional/insomnia triggers, motion sickness history, otoacoustic emissions) achieved an AUC of 0.8788.

Conclusions:

• The developed multimodal model demonstrates high preliminary diagnostic accuracy for VM.
• This predictive tool has the potential to improve clinical diagnosis of vestibular migraine.
• Further validation in larger, prospective studies is necessary to confirm generalizability.