AI in Neurological Practice: From Ideas to First Results — Based on EAN 2026

This review summarizes six presentations from two scientific sessions at the European Academy of Neurology (EAN) 2026 congress devoted to the application of artificial intelligence (AI) in outpatient and inpatient neurology: epileptology, movement disorders, headache medicine, vascular neurology, disorders of consciousness and demyelinating disease. Drawing on talks by Sándor Beniczky, Christian Lambert, Anker Stubberud, Susanne Wegener, Anna Estraneo, and Celia Oreja-Guevara, the article identifies a recurring pattern across all six domains: a substantial gap persists between the number of published algorithms claiming «expert-level» accuracy and the number of tools that have obtained regulatory approval (CE marking, FDA clearance) and are actually deployed in routine practice. The review maps which AI applications already have certified, ready-to-use solutions (automated interictal epileptiform discharge detection on electroencephalography, wearable seizure detectors, quantitative magnetic resonance imaging and optical coherence tomography analysis in multiple sclerosis) and which remain at the research stage (probabilistic diagnosis of parkinsonism, prediction of headache treatment response, detection of covert consciousness, post-stroke outcome prediction). It also discusses shared methodological caveats — circular logic in diagnostic models, optimism bias among human experts, and the consistent advantage of hybrid «algorithm plus clinician» workflows over full automation — and offers practical recommendations for Ukrainian neurologists on evaluating, adopting, and verifying the local availability of these technologies.