Table 1.
Performance of pretrained transformer language models in detecting patients with ADRD
Language model | Precision | Recall | F1-score | Accuracy | AUC-ROC |
---|---|---|---|---|---|
BERT base-cased | 75.67 | 80 | 77.77 | 77.46 | 83.73 |
DistilBERT | 84.84 | 80 | 82.35 | 83.09 | 89.05 |
XLNET | 73.68 | 80 | 76.71 | 76.05 | 83.17 |
DistilRoBERTa | 79.41 | 77.14 | 78.26 | 79.20 | 78.87 |