The study protocol is illustrated in Figure 1. The sample consisted of 30 abstinent individuals with past diagnosis of AUD and 30 healthy volunteers (CTL) (see Section 2.1 for details). The analytic measures included rs-fMRI (Section 2.4, Section 2.5 and Section 2.6), self-rated impulsivity scores (Section 2.3), and neuropsychological test scores (Section 2.2). Major analyses employed in the study were (i) feature selection to extract relevant features that will be used in classification analysis (Section 2.7), (ii) random forest method to identify the key features that significantly contribute to classifying AUD from the CTL participants (Section 2.8), (iii) zero-order correlations were used to compute associations of the key variables identified by the random forest classification model, (a) among themselves (Section 3.2) and (b) with age in each group (Section 3.3). Partial correlations were employed to identify associations between age and the key variables of classification by controlling the group effects in the total sample (Section 3.3).
