This project aims to find differences in EEG microstates between patients with alcohol use disorder (AUD) and healthy control (HC) using machine learning models. EEG microstates are inferred from EEG recordings during resting states. The project identifies four microstates and constructs a connectivity matrix for each microstate informed by a coherence analysis. The upper triangle portion of the connectivity matrix is used as input to SVM classification model that discriminates between the AUD and HC groups. Among four microstates, the classification accuracy reaches the maximum of 96.7% with the data from a particular microstate (microstate (MS) D in Fig. 4).