Machine Learning-Based Classification of Disorders of Consciousness Using Multi-Dimensional EEG Features

Disorders of Consciousness (DOC), including Vegetative State (VS) and Minimally Conscious State (MCS), present significant diagnostic challenges owing to the inherent subjectivity of behavioral assessments. Electroencephalography (EEG) provides a noninvasive and objective modality for evaluation. This study proposes a multi-dimensional EEG feature extraction framework encompassing temporal, spectral, and spatial domains. To classify VS and MCS, machine learning models including Support Vector Machine (SVM), Logistic Regression (LR), Random Forest (RF), and Convolutional Neural Network (CNN) were employed. Experimental results demonstrate that the SVM model achieved the highest performance, with a classification accuracy of 93% and an F1 score of 96%. Slow-wave and variability-related features were identified as the most significant contributors to model performance. These findings underscore the potential of multi-dimensional EEG features to enhance diagnostic accuracy, offering substantial clinical value for improving patient management and therapeutic outcomes.