Fig. 1: Area under the curve (AUC) for machine learning models discriminating between female patients with NSSI + SA and those with NSSI alone.

Panel (A) includes the reduced set and panel (B) includes the full set. The bars represent the performance of machine learning models: A model with AUC = 1 would be considered as perfect, 0.9 < AUC < 1 as excellent, 0.8 < AUC < 0.9 as good, 0.7 < AUC < 0.8 as fair, 0.6 < AUC < 0.7 as poor and 0.5 < AUC < 0.6 as being slightly better than chance. The error bars represent the 95% confidence intervals (CIs) of AUC values.