Machine learning model performance in identifying patients with PASC (MIS-C or non-MIS-C variants).

The table displays several performance metrics for each of the three outcomes (PASC (any), non MIS-C PASC, and MIS-C). Accuracy, F1 score, precision, and recall were all computed at the p = 0.5 threshold. In addition to composite performance statistics, we report the area under the precision-recall curve (AUPR) and area under the receiver operating characteristic curve (AUROC). The curves themselves are shown and described in Fig 2.