Skip to main content

Table 4 Performances of the prediction model and scoring systems

From: Development and validation of a survival prediction model in elder patients with community-acquired pneumonia: a MIMIC-population-based study

Indicators

PPV (95% CI)

NPV (95% CI)

AUC (95% CI)

ACC (95% CI)

Training set

 Prediction model

0.529 (0.433–0.626)

0.850 (0.810–0.890)

0.751 (0.749–0.752)

0.769 (0.727–0.810)

 SOFA

0.448 (0.348–0.547)

0.817 (0.774–0.860)

0.672 (0.670–0.674)*

0.729 (0.685–0.772)

 SOAR

0.302 (0.242–0.362)

0.825 (0.769–0.881)

0.607 (0.605–0.609)*

0.532 (0.484–0.581)

 PSI

0.273 (0.225–0.321)

0.875 (0.799–0.951)

0.538 (0.536–0.540)*

0.381 (0.333–0.428)

 CURB-65

0.332 (0.267–0.396)

0.843(0.792–0.893)

0.645 (0.643–0.646)*

0.582 (0.534–0.630)

Testing set

    

 Prediction model

0.475 (0.350–0.601)

0.827 (0.768–0.886)

0.744 (0.742–0.747)

0.728 (0.669–0.787)

 SOFA

0.356 (0.216–0.495)

0.767 (0.704–0.831)

0.634 (0.632–0.637)*

0.682 (0.620–0.744)

 SOAR

0.360 (0.271–0.450)

0.849 (0.781–0.917)

0.667 (0.665–0.670)

0.599 (0.534–0.664)

 PSI

0.301 (0.233–0.369)

0.927 (0.847–1.000)

0.621 (0.618–0.623)*

0.419 (0.354–0.485)

 CURB

0.538 (0.347–0.730)

0.780 (0.721–0.839)

0.671 (0.668–0.673)*

0.751 (0.694–0.809)

External validation

 Prediction model

0.467 (0.418–0.516)

0.790 (0.764–0.817)

0.703 (0.672–0.734)

0.692 (0.666–0.717)

 SOFA

0.448 (0.403–0.494)

0.801 (0.774–0.828)

0.686 (0.654–0.718)

0.675 (0.649–0.701)

 SOAR

0.327 (0.294–0.361)*

0.766 (0.731–0.802)

0.564 (0.533–0.594)*

0.511 (0.484–0.539)*

 PSI

0.379 (0.335–0.423)*

0.762 (0.734–0.791)

0.613 (0.580–0.646)*

0.626 (0.599–0.653)*

 CURB-65

0.329 (0.295–0.362)*

0.769 (0.734–0.805)

0.570 (0.538–0.601)*

0.511 (0.484–0.539)*

  1. PPV, positive predictive value; NPV, negative predictive value; AUC, area under the curve; ACC, accuracy; CI: confidence interval; SOFA, Sequential Organ Failure Assessment; SOAR, systolic blood pressure, oxygenation, age and respiratory rate; PSI, Pneumonia Severity Index; * is statistically significant compared with the prediction model