Table 3 Comparison between the state-of-the-art methods and our framework
From: Leveraging text skeleton for de-identification of electronic medical records
Model | 2006 i2b2 | 2014 i2b2 | Chinese | ||||||
|---|---|---|---|---|---|---|---|---|---|
Precision | Recall | F1-score | Precision | Recall | F1-score | Precision | Recall | F1-score | |
Wellner | 0.9870 | 0.9750 | 0.9810 | – | – | – | – | – | – |
Nottingham | – | – | – | 0.9900 | 0.9640 | 0.9768 | – | – | – |
MIST | – | – | – | 0.9529 | 0.7569 | 0.84367 | – | – | – |
CRF | 0.9640 | 0.9371 | 0.9504 | 0.9842 | 0.9663 | 0.9752 | 0.9863 | 0.9705 | 0.9783 |
CRF + ANN | – | – | – | 0.9792 | 0.9784 | 0.9788 | – | – | – |
Bi-LSTM | 0.9723 | 0.9656 | 0.9689 | 0.9878 | 0.9389 | 0.9627 | 0.9908 | 0.9584 | 0.9743 |
Bi-GRU | 0.9871 | 0.9664 | 0.9766 | 0.9750 | 0.9704 | 0.9727 | 0.9898 | 0.9624 | 0.9759 |
TS-GRU | 0.9903 | 0.9855 | 0.9879 | 0.9889 | 0.9723 | 0.9805 | 0.9875 | 0.9719 | 0.9797 |