It is evident from the ROC curves shown in Figure 1a that density measures present greater diagnostic accuracy at classifying AUD diagnosis compared to the ROC curves for dichotomous measures (Figure 1b). As shown in Table 2b, although AUC for all eight FH measures were significantly greater than 50% probability, AUC for all dichotomous measures were closer to 50% probability, indicating poor diagnostic accuracy. FHFirst–AUDDx and FHDrPSND–AUDMaxSxCnt had the highest AUC among dichotomous and density measures respectively. Overall, density (vs. dichotomous) measures presented higher AUC indicative of better predictive accuracy at classifying AUD diagnosis.
