Table 2.

Performance of the ensemble Super Learner in the prediction of postacute sequelae of COVID-19 diagnosis. Model created using electronic health record data from a sample of patients included in the National COVID Cohort Collaborative during the COVID-19 pandemic.

Metric	Estimate (95% CI)
Area under the receiver operator curve	0.874 (0.864-0.884)
Accuracy	0.772 (0.761-0.783)
Precision	0.467 (0.446-0.489)
Recall	0.806 (0.784-0.828)
F1-score	0.591 (0.571-0.661)
Brier score	0.110 (0.104-0.116)