Table 3.
Detailed performance for each SBDoH category for the best NLP model - BERT_general.
BERT_general | Strict | Lenient | ||||
Type | Precision | Recall | F1-score | Precision | Recall | F1-score |
Gender | 0.9091 | 0.9167 | 0.9129 | 0.9091 | 0.9167 | 0.9129 |
Ethnicity | 0.8571 | 1.0000 | 0.9231 | 0.8571 | 1.0000 | 0.9231 |
Education | 0.8857 | 0.8493 | 0.8671 | 0.9286 | 0.8904 | 0.9091 |
Smoking | 0.8764 | 0.8764 | 0.8764 | 0.9045 | 0.9045 | 0.9045 |
Employment | 0.6667 | 0.4000 | 0.5000 | 0.6667 | 0.4000 | 0.5000 |
Overall | 0.8848 | 0.8734 | 0.8791 | 0.9058 | 0.8941 | 0.8999 |