Assess the level of consciousness in patients with disorders of consciousness by combining resting-state and auditory-evoked EEG.

Introduction: Electroencephalography (EEG) can provide objective neural marker for assessing the level of consciousness of patients with disorders of consciousness (DoC), but current research mainly focuses on the EEG features of a single modality, such as the resting-state or the evoked state, which results in less than ideal assessment accuracy. To accurately assess the level of consciousness of DoC patients, we proposed a new method by combine with resting-state and auditory-evoked EEG. Methods: The EEG data of resting-state and auditory-evoked potential were collected from 157 DoC patients. Then, nonlinear dynamics feature (NDF) include spatiotemporal correlation entropy and neuromodulation intensity of multimodal EEG were extracted. Next, the multi-form feature selection algorithm (MFFS) was adopted to optimize the extracted EEG features. Finally, a diagnosis model was constructed using support vector machine (SVM). Results: Among them, SC-Theta, SC-Alpha, NI-Alpha and ERP features were significantly ( Discussion: The proposed diagnostic model has demonstrated its distinctive advantages, showcasing remarkable effectiveness and reliability in accurately assessing consciousness states. This method holds promise for improving the reliability of clinical awareness assessments.