EUVIMEDEUVIMED entdecken
Crossref

Directed Chebyshev Spectral Graph Neural Networks: Continuous MoCA Prediction from EEG in Parkinson’s Disease

Ahmed Afandy, Amin Mohammed, Moayad Potrus

Sulaimani Journal for Engineering Sciences · 01.08.2026 · 12 (3) · 153-188

DOI: 10.17656/sjes.10222 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sulaimani Journal for Engineering Sciences, online erschienen am 01.08.2026. Verfasst von Ahmed Afandy und 2 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Cognitive impairment is a significant and debilitating feature of Parkinson’s disease (PD) among its non-motor symptoms. However, the current clinical tools for assessing cognitive impairment, such as the Montreal Cognitive Assessment (MoCA) test, are subject to several limitations. To address these issues, we developed a deep learning model for predicting the cognitive status of patients with Parkinson’s disease objectively and continuously from their resting-state electroencephalography (EEG) signals. We model the EEG signals as directed brain networks in the frequency domain using Partial Directed Coherence (PDC) and then process these network representations using a graph neural network with Chebyshev spectral convolutions and a self-attention mechanism. We rigorously tested our model using a dataset from 83 patients with Parkinson’s disease and compared its performance with the state-of-the-art deep learning models. Our model outperformed the state-of-the-art models by reaching the highest performance in the Theta (4-8 Hz) and Alpha (8-12 Hz) bands. At the subject level, the model achieved a mean absolute error (MAE) of 0.34 points (0-30 MoCA) and R² of 0.98. These results suggest that the directed network topology of the EEG signals is a sensitive and physiologically meaningful feature for the prediction of cognitive impairment in Parkinson’s disease.

Zitieren

APA

Afandy, A., Mohammed, A., & Potrus, M. (2026). Directed Chebyshev Spectral Graph Neural Networks: Continuous MoCA Prediction from EEG in Parkinson’s Disease. Sulaimani Journal for Engineering Sciences, 12(3), 153-188. https://doi.org/10.17656/sjes.10222

Vancouver

Afandy A, Mohammed A, Potrus M. Directed Chebyshev Spectral Graph Neural Networks: Continuous MoCA Prediction from EEG in Parkinson’s Disease. Sulaimani Journal for Engineering Sciences. 2026;12(3):153-188. doi:10.17656/sjes.10222

BibTeX
@article{Afandy2026,
  author = {Afandy, Ahmed and Mohammed, Amin and Potrus, Moayad},
  title = {Directed Chebyshev Spectral Graph Neural Networks: Continuous MoCA Prediction from EEG in Parkinson’s Disease},
  journal = {Sulaimani Journal for Engineering Sciences},
  year = {2026},
  volume = {12},
  number = {3},
  pages = {153--188},
  doi = {10.17656/sjes.10222}
}
Kostenlos starten →