Table 4 Performance comparison in test set.

From: Development of early prediction model for pregnancy-associated hypertension with graph-based semi-supervised learning

Models

AUROC

Sensitivity

Specificity

PPV

NPV

\({\mathrm{SSL}}_{\mathrm{L}+\mathrm{U}}^{[\mathrm{b}]}\)

0.811

0.727

0.800

0.099

0.989

\({\mathrm{LR}}^{[\mathrm{b}]}\)

0.762

0.636

0.663

0.054

0.984

\({\mathrm{SVM}}^{[\mathrm{b}]}\)

0.701

0.719

0.795

0.096

0.989

\({\mathrm{RF}}^{[\mathrm{b}]}\)

0.725

0.725

0.666

0.062

0.987

Risk factors

–

0.455

0.805

0.066

0.980

  1. Risk factors: conventional risk factors recommended by American College of Obstetricians and Gynecologists.
  2. AUROC area under the ROC curve, PPV positive predicted value, NPV negative predicted value; [a]: models with variables from clinical guidelines, [b] models with selected important variables, and [c] models with all routine variables.
  3. The performances of the best model are in bold.
Back to article page