Alcoholic EEG Data Classification Using Weighted Graph-Based Technique

The analysis and classification of Alcohol Use Disorder (AUD) using non-invasive measurements, such as EEG records from the brain scalp, are of significant importance in neuroscience. Analysis and diagnosis of brain diseases associated with alcoholic subjects using EEG records remain challenging. This study proposes a graph theory-based approach for automated classification of AUD using EEG data. The metrics of the graphs are intrinsically related to the organization of the brain functionality. The main contribution of this study is to evaluate the impact of weighted graph features on AUD classifications based on EEG data. In this study, three different features (average degree, fluctuation difference, and average weighted degree) were extracted from the weighed visibility EEG graph and the performance of the proposed model was evaluated against SVM, k-NN, and Naive Bayes classifiers. The experimental results indicates that the topological features of the weighted EEG graphs supported superior classification performance (97.5%) against the other competing methods.