EEG-based Machine Learning Models for Early Detection of Dementia with Lewy Bodies in Isolated REM Sleep Behavior Disorder: A Scoping Review
Mahmoud Hefnawy1, Ahmed Negida2
1Zagazig University Faculty Of Medicine, 2Virginia Commonwealth University
Objective:
This scoping review aims to summarize the current evidence on EEG-derived and AI-enhanced models for the early prediction of DLB based on neurophysiological patterns observed in patients with iRBD.
Background:
Isolated rapid eye movement sleep behavior disorder (iRBD) is recently recognized as a prodromal stage of α-synucleinopathies, particularly dementia with Lewy bodies (DLB). Quantitative electroencephalography (EEG) and artificial intelligence (AI) based analyses have recently emerged as promising tools for detecting early cortical dysfunction and predicting phenoconversion to DLB.
Design/Methods:
This scoping review was conducted in accordance with PRISMA-ScR guidelines. A systematic search of PubMed, Scopus, Web of Science, and the Cochrane Library was performed from inception to July 2025. Studies enrolling polysomnography-confirmed iRBD or prodromal DLB with REM sleep behavior disorder that analyzed EEG features or AI-based EEG models were included. Data on EEG parameters, spectral analysis, and diagnostic performance were extracted.
Results:
Four studies met the criteria, including over 500 participants with iRBD, prodromal DLB, and controls. Across studies, iRBD showed global EEG slowing with increased theta–delta power and reduced posterior alpha activity, most prominently in parieto-occipital regions. Phenoconversion to DLB was associated with a further reduction in dominant alpha frequency, disrupted posterior connectivity, and increased network randomness on functional connectivity analysis. Machine-learning and deep-learning models trained on EEG spectrograms or connectivity matrices achieved AUCs of 0.83–0.89 for predicting conversion.
Conclusions:
EEG alterations in iRBD align with those observed in prodromal DLB, characterized by cortical slowing and disrupted posterior connectivity as early electrophysiological markers. AI-driven EEG analysis offers a promising, non-invasive approach for identifying individuals at risk of developing DLB.
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.