Table 1.
Descriptive statistics.
Fig 1.
Multilayer stacked ensemble (the shaded boxes are learned).
Adapted from [46].
Table 2.
ROC AUC performance for the structured data.
Table 3.
Performance metrics for the weighted ensemble model.
Fig 2.
Permutation feature importance for the weighted ensemble model.
Table 4.
Performance of bio clinical BERT on the unstructured data.
Fig 3.
Averaged local (LIME) absolute values feature importance for the BERT Transformer (left: lemmatization, right: steming).
Table 5.
Performance of the best models for mixed data.
Table 6.
Permutation feature importance for the best models on mixed data.