Model performance metrics.

Model performance was evaluated using F1 score, Negative Predictive Value (NPV), Positive Predictive Value (PPV), Sensitivity/Recall, and Specificity. The F1 score represents the harmonic mean of the Precision and Recall; NPV is a calculation of true negatives / all negatives; PPV is a calculation of true positives / all positives; Sensitivity/Recall is a calculation of true positives / (all true positives and false negatives); and Specificity is a calculation of true negatives / (all true negatives and false positives). TN, FN, TP, and FP refer to true negatives, false negatives, true positives, and false positives respectfully.