Table 5 CRF papers on de-identification of medical free text
From: Survey on RNN and CRF models for de-identification of medical free text
Author | F-measurea | Precisiona | Recalla |
|---|---|---|---|
Berg and Dalianis, 2019 [71] | 91.00 | 94.66 | 86.72 |
Bui et al. 2017 [72] | 93.66 | 96.29 | 91.18 |
Bui et al. 2018 [73] | 95.10 | 98.50 | 92.00 |
Du et al. 2018 [74] | 98.78 | 99.27 | 98.29 |
Kajiyama et al. 2018 [39] | 80.61 | Not provided | Not provided |
Kim et al. 2018 [40] | 95.73 | 97.04 | 94.45 |
Lee et al. 2017 [75] | 90.74 | 93.39 | 88.30 |
Lee et al. 2017 [76] | 90.40 | 93.46 | 87.53 |
Liu et al. 2017 [43] | 96.98 | 97.94 | 96.04 |
Phuong et al. 2016 [77] | 96.00 | 97.91 | 94.16 |
Trienes et al. 2020 [47] | 91.20 | 95.90 | 86.90 |
Yang et al. 2019 [48] | 96.46 | 97.97 | 94.98 |