The purpose of this study is to prove that virtual reality technology can be utilized to improve the reliabilty and accuracy of components of the neurologic examination compared to remote telemedicine encounter.
We plan to build a prototype device using existing virtual reality (VR) headset, the Pico Neo 3 Pro Eye. This system incorporates eye and limb tracking technology that will enable us to develop software programs/modules that allow physicians to assess components of the neurologic examination including level of consciousness, extraocular movements, gaze preference, visual fields, visual neglect, visual acuity, pupillary response, arm drift, ataxia, language (aphasia), and speech (dysarthria). We plan to test the performance of the VR headset examination against both the in-person bedside National Institutes of Health Stroke Scale (NIHSS) and the telemedicine NIHSS examination.
Research subjects will include prospectively identified patients admitted to the hospital in the setting of acute stroke with last known well time within 7 days. We plan to test the performance of the VR headset examination against both the in-person bedside neurologic examination and audio-visual telemedicine examination. Research subjects will be tested in series using all three testing platforms and then interrater reliability of the VR headset will be evaluated by comparing to the other two validated testing modalities.
As part of the study, data collection from medical record will include age, gender, ethnicity, preferred language, handedness (left or right), baseline admission NIHSS, stroke subtype (ischemic or hemorrhagic), stroke location (dominant or nondominant hemisphere), and if ischemic—stroke mechanism (lacunar vs nonlacunar), anterior or posterior circulation, and (if applicable) arterial territory. Additionally, participants will be asked to fill out a questionnaire following completion of the VR assessment to assess subjective measures including device comfort, quality of testing, and user friendliness.
Following validation of the VR testing modules, we plan to use the anonymized data from the VR headsets to develop supervised machine learning models that can automatically score components of the neurologic assessment.