Parkinson's EEG classification model based on the channel optimization

In recent years, Electroencephalogram (EEG) has gradually become one of the important means of early Parkinson's disease (PD) detection because of its non-invasive and real-time monitoring capability. This paper discusses the EEG signal analysis method from the effective feature extraction techniques and the application of traditional machine learning algorithms. Although traditional methods are able to provide some degree of feature interpretability, there are limitations in handling high-dimensional feature space and redundant features, resulting in decreased classification performance. Therefore, this paper proposes a Support Vector Machine (SVM) classification model based on artificial channel optimization, through channel selection of EEG signals recorded by 64-lead devices, screening out the optimal channel combination highly related to PD, and then achieving efficient classification. Specifically, the study uses publicly available Parkinson's disease EEG datasets, employs power spectral density (PSD) features, and combines manual channel selection with the strong classification capabilities of SVM to achieve effective classification of PD EEG signals. In addition, the article highlights that model performance may vary on different datasets, and more clinical validation is needed in the future, and suggests to combine other machine learning technologies such as deep learning to further improve model performance.