Recall severity prediction on FDA recall data (test)
96.3AccuracyLightGBM
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| LightGBMInput=GloVe + Tabular, Model category=Boosting2026.06 | 96.3 | 79.4 | 95.6 | 85.6 | 97.4 | |
| XGBoostInput=GloVe + Tabular, Model category=Boosting2026.06 | 96 | 79.3 | 93.8 | 85 | 96.7 | |
| RecallRisk-BERTInput=Text + Tabular, Model category=Proposed2026.06 | 95.1 | 84.9 | 83.3 | 84.1 | 95.8 | |
| Linear SVMInput=TF-IDF + Tabular, Model category=Classical ML2026.06 | 94.5 | 82.2 | 81.1 | 81.6 | 93.7 | |
| BiLSTM + AttentionInput=GloVe + Tabular, Model category=Deep learning2026.06 | 93.3 | 85.4 | 77.2 | 80.7 | 95.4 | |
| DNNInput=GloVe + Tabular, Model category=Deep learning2026.06 | 92.8 | 83.3 | 75.6 | 79.1 | 94.8 | |
| PubMedBERTInput=Text + Tabular, Model category=Transformer2026.06 | 90.2 | 84.7 | 73.2 | 76.8 | 95.4 | |
| Logistic RegressionInput=TF-IDF + Tabular, Model category=Classical ML2026.06 | 89.5 | 84.1 | 67.8 | 73.9 | 94.8 |