Figure legends
Figure 1. Receiver operating characteristic (ROC) of machine
learning prediction models
Figure 1A: Antepartum prediction of massive PAS-associated blood
loss - ROC of train (left) and test sets (right)Figure 1B: Antepartum prediction of prolonged hospitalization -
ROC of train (left) and test sets (right)Figure 1C: Antepartum prediction of admission to intensive care
unit - ROC of train (left) and test sets (right)Figure 1D: Peripartum prediction of massive PAS-associated blood
loss - ROC of train (left) and test sets (right)Figure 1E: Peripartum prediction of prolonged hospitalization -
ROC of train (left) and test sets (right)Figure 1F: Peripartum prediction of admission to intensive care
unit - ROC of train (left) and test sets (right)
Figure 2. Size of contribution of baseline, disease, and
management variables in antepartum and peripartum prediction models
(variables are ordered in a clockwise direction starting from 12
o’clock)
Figure 2A: Antepartum prediction of massive PAS-associated blood
loss - ROC of train (left) and test sets (right)Figure 2B: Variables contributing to antepartum prediction of
prolonged hospitalizationFigure 2C: Variables contributing to antepartum prediction of
admission to intensive care unitFigure 2D: Variables contributing to peripartum prediction of
massive PAS-associated blood lossFigure 2E: Variables contributing to peripartum prediction of
prolonged hospitalizationFigure 2F: Variables contributing to peripartum prediction of
admission to intensive care unit
Figure 3. A “box and whisker” graph of calculated probability
of adverse outcomes in women
who developed or did not develop these outcomes
Figure 3A: Antepartum probability of massive PAS-associated blood
lossFigure 3B: Antepartum probability of prolonged hospitalizationFigure 3C: Antepartum probability of admission to intensive care
unitFigure 3D: Peripartum probability of massive PAS-associated blood
lossFigure 3E: Peripartum probability of prolonged hospitalizationFigure 3F: Peripartum probability of admission to intensive care
unit