Conversational Large Language Models for Vestibular Diagnosis in Outpatient Clinics: Prospective Multicenter Diagnostic Accuracy Study

Vestibular disorders are common, burdensome, and frequently misdiagnosed, particularly in nonspecialist settings where history-taking is often incomplete or inconsistently structured. Digital health tools that standardize symptom elicitation could improve diagnostic triage, but most existing systems rely on static questionnaires or rule-based logic....

Continue reading at PubMed →

Results are preliminary and from a single prospective study; do not integrate conversational LLMs into vestibular diagnostic workflows without further validation showing clinical-grade accuracy and safety.

Vestibular misdiagnosis is a persistent clinical problem, and rigorous prospective evidence on whether AI chatbots can close that gap has direct implications for triage and referral pathways in audiology and otolaryngology.

1. 01Prospective, multicenter design adds methodological rigor compared with retrospective LLM benchmarking studies.
2. 02Conversational LLMs were evaluated specifically for vestibular disorder diagnosis — an area with high misdiagnosis rates.
3. 03Outpatient clinic setting reflects real-world conditions rather than controlled lab or vignette-only testing.
4. 04Diagnostic accuracy was the primary outcome, making results directly interpretable for clinicians.
5. 05Findings could inform whether AI tools warrant integration into balance-disorder triage or referral processes.

PMID

42572244

DOI

10.2196/100442.

Journal

Journal of Medical Internet Research

Publication type

research_article

Evidence level

2b

Population

Patients presenting with dizziness or balance complaints at outpatient clinics across multiple centers

Intervention

Conversational large language models used for vestibular disorder diagnosis

Comparator

Clinician diagnosis or reference standard vestibular diagnosis