Despite being common, disabling, and treatable, vertigo remains undertreated globally due to limited expertise, particularly in emergency departments (ED). Untreated vertigo leads to falls, injuries, loss of livelihood and premature death. Vertigo is characterized by a distinctive history and eye-movement abnormalities, indicating the underlying vestibular disorder. This proposal aims to develop, validate, and deploy an Artificial Intelligence-powered "Electronic Vestibular Assistant (EVA)", the first virtual expert system for vertigo.
Healthcare workers using EVA will first input patient history digitally, receive a preliminary "syndrome diagnosis" and recommendations for specific eye examinations, such as the Hallpike Test for Benign Positional Vertigo and the Head Impulse Test for Vestibular Neuritis. An automated nystagmus detector and head impulse tool will aid in performing and interpreting these examinations. A machine learning algorithm will provide the likelihood of specific diagnoses coupled with treatment options. Following development and validation, we will compare EVMs accuracy with that of real-life experts and assess its impact in ED compared to standard care.