Skip to main content
. 2021 Jul 2;36(31):e198. doi: 10.3346/jkms.2021.36.e198

Table 2. Performance of the machine learning model for each health outcome of interest.

Dataset Metric Anaphylaxis performance Agranulocytosis performance
Training set Recall 0.940 0.949
Precision 0.945 0.953
f1-score 0.939 0.949
Accuracy 0.940 0.949
AUROC 0.995 0.994
Test set Recall 0.698 0.729
Precision 0.695 0.727
f1-score 0.670 0.726
Accuracy 0.698 0.729
AUROC 0.697 0.770

AUROC = area under the receiver operating curve.