0.001 (Figure 3). Overall, out of all the combined models of EEG+SNP features, the AA & EA female groups achieved the highest accuracy of 79.33% (specificity=71.02%, sensitivity = 87.67%, AUC=0.99, F=0.81), 78.91% (specificity=76.82%, sensitivity = 81%, AUC=0.9, F=0.79), respectively, and the AA & EA early adolescence range age of 79.54% (specificity=79.55%, sensitivity = 79.52%, AUC=0.93, F=0.79), 74.2% (specificity=68.43%, sensitivity = 79.23%, AUC=0.89, F=0.76), respectively (full list of the significant models in Table 1). Interestingly, we found gender and ethnicity differences when comparing the addition of the FH feature of mother DSM-5 AUD or father DSM-5 AUD to the combined model of EEG+SNP. For both AA and EA, male and female samples, mother AUD feature increased model accuracy EA:(p(son-mother vs. EEG+SNP)<0.001) (p(daughter -mother vs. EEG+SNP)=0.02), AA:(p(son-mother vs. EEG+SNP)=0.001, p(daughter -mother vs EEG+SNP)<0.001). Father AUD increased the accuracy of the combined model only for the AA female sample (p (daughter-father vs EEG+SNP) <0.001) (Table 1, Figure 4). Finally, the AA female group with the combined model of EEG+SNP features with the addition of FH of father AUD or mother AUD feature achieved the highest accuracy of 87.55% (father AUD) (specificity =85.71%, sensitivity=89.38%, AUC=0.99, F=0.89) and 87.11% (mother AUD)(specificity =81.3%, sensitivity=92.92%, AUC=0.99, F=0.88).
