Artificial intelligence for epilepsy decision support.
Publication/Presentation Date
11-21-2025
Abstract
Artificial intelligence (AI) is increasingly being applied in epilepsy, a field marked by diagnostic uncertainty, heterogeneous treatments, and variable outcomes. This review examines the current and potential roles of AI in key domains including history-taking, electronic records, electroencephalography (EEG), magnetic resonance imaging (MRI), and multimodal integration. Three opportunities are highlighted: efficiency gains through task automation; improvement in diagnostic accuracy, including outcome-based validation; and treatment applications such as prediction of risk of seizure recurrence, drug resistance, and surgical candidacy. We review the challenges when measuring outcomes of pharmacological, surgical, and device therapies. Workflow integration and the potential of AI to reduce administrative burden are also discussed. Ethical concerns, including bias, transparency, accountability, privacy, sustainability, and risks of overreliance are reviewed as critical to responsible adoption. Ultimately, AI holds promise as a decision-support tool that may enhance, but not replace, clinical judgment, provided it is validated in diverse real-world settings and developed in partnership with clinicians.
ISSN
1528-1167
Published In/Presented At
Loushy, I., & Sperling, M. R. (2025). Artificial intelligence for epilepsy decision support. Epilepsia, 10.1111/epi.18709. Advance online publication. https://doi.org/10.1111/epi.18709
Disciplines
Medicine and Health Sciences
PubMedID
41269050
Department(s)
Department of Medicine
Document Type
Article