Table 5. Comparison of the recall of different BERT (Bidirectional Encoder Representations from Transformers) models.
| Model size and model name | All codes, 95% CI | Top 80% codes, 95% CI | ||
|---|---|---|---|---|
| Multilabel | Top-5 | Multilabel | Top-5 | |
| Base | ||||
| NorDeClin-BERT-base-NorICDa | 0.45 (0.44-0.46) | 0.81 (0.80-0.81) | 0.51 (0.50-0.52) | 0.87 (0.86-0.88) |
| SweDeClin-BERT-SweICD | 0.25 (0.24-0.26) | 0.59 (0.58-0.60) | 0.29 (0.28-0.30) | 0.65 (0.63-0.66) |
| SweDeClin-BERT-NorICD | 0.41 (0.40-0.42) | 0.78 (0.77-0.79) | 0.48 (0.47-0.49) | 0.85 (0.84-0.86) |
| ScandiBERT-NorICD | 0.40 (0.39-0.41) | 0.78 (0.77-0.79) | 0.48 (0.47-0.49) | 0.85 (0.84-0.86) |
| NorBERT3-base-NorICD | 0.44 (0.43-0.45) | 0.80 (0.79-0.81) | 0.51 (0.50-0.52) | 0.86 (0.86-0.87) |
| Large | ||||
| NorDeClin-BERT-large-NorICD | 0.48 (0.47-0.49)b | 0.82 (0.82-0.83)b | 0.54 (0.53-0.55)b | 0.88 (0.88-0.89)b |
| NorBERT3-large-NorICD | 0.43 (0.42-0.44) | 0.81 (0.80-0.82) | 0.50 (0.49-0.51) | 0.88 (0.87-0.88)b |
ICD-10: International Statistical Classification of Diseases, Tenth Revision.
Highest score for each scenario.