Work

Subject-Wise WST–AlexNet Framework for EEG-Based Parkinson's Disease Classification and Channel Screening

Year: 2026

Type: conference-paper

Abstract: Reliable subject-level model assessment is important for EEG-based Parkinson’s disease (PD) classification in health-informatics research. This paper presents a subject-wise wavelet scattering transform (WST)–AlexNet framework for EEG-PD classification and candidate-channel screening using Iowa and UCSD/San Diego EEG datasets. The targeted leave-one-subject-out (LOSO) evaluation of benchmark-selected AFz/AF4 channels achieved 67.86% accuracy for Iowa AFz, 58.06% for UCSD PD-OFF AF4, and 48.39% for UCSD PD-ON AF4. Exploratory all-channel screening identified AFz, Oz, and F8 as stronger candidate channels, with observed subject-level accuracies of 75.00%, 74.19%, and 70.97%, respectively. The framework provides a compact health-informatics pipeline for subject-wise EEG-PD model assessment, benchmark interpretation, and candidate-channel discovery.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.21970239

Authors: Akbari, Hesam, Niloofar, S. Alizadeh, Sadiq, Muhammad Tariq, Siuly, Siuly, Mete, Mutlu

Institutions: University of Essex, Victoria University, University of North Texas

Language: English

Topic: EEG and Brain-Computer Interfaces

Subfield: Cognitive Neuroscience

Field: Neuroscience

Domain: Life Sciences

SDG: Good health and well-being

Open Access status: green

OpenAlex ID: W7203576547

FWCI:

Cites:

Cited by:

Related to: 10