Table 3 Performance metrics of different XGBoost models for predicting the given record as COVID-19-positive or influenza or COVID-19-negative when tested on internal validation or test set.

From: A vital sign-based prediction algorithm for differentiating COVID-19 versus seasonal influenza in hospitalized patients

Performance measure

COVID-19-positive vs influenza

Influenza vs others

Covid-19-positive vs COVID-19-negative

AUC

0.988

0.986

0.973

Accuracy

0.949

0.945

0.930

F1 score

0.928

0.918

0.878

Sensitivity (recall)

0.925

0.913

0.872

Specificity

0.962

0.961

0.954

PPV (precision)

0.932

0.924

0.884

NPV

0.958

0.955

0.945

FPR

0.038

0.039

0.046

FDR

0.069

0.076

0.116

FNR

0.075

0.087

0.128

Back to article page