The RF classification model included 66 DMN connections for each of the 5 frequency bands, 13 neuropsychological scores, and 4 BIS scores as features, while the group status (AUD and CTL) served as the outcome variable. The training data consisted of full sample for identifying significant features for classifying the groups. To compute prediction error and classification accuracy, we used the out-of-bag (OOB) error estimate, which represents classification error obtained from the out of bag sample that were not part of the bootstrap sample used in growing the forests. In RF model, cross-validation in a separate test sample is not required, as it is estimated internally in the algorithm [107]. During each iteration of constructing a decision tree, only about two-thirds of the bootstrap sample from the training data is used and about one-third of the sample is left out during each bootstrap process, which is called the out-of-bag (OOB) sample. The classification error calculated from this sample is called the OOB error score. The aggregate of OOB scores from all decision trees will provide the ensemble OOB error rate
