Skip to main content

Table 5 Evaluation metrics obtained with LSTM-LSTM-CRF with EMA model for each named entity

From: De-identifying Spanish medical texts - named entity recognition applied to radiology reports

  Training set Validation set Test set MEDDOCAN
  Precision Recall F1 Precision Recall F1 Precision Recall F1 Precision Recall F1
CAB 98.29 97.94 98.11 96.03 94.92 95.47 83.69 75.76 79.53 13.33 33.33 19.05
DIR 100 100 100 93.49 95.00 94.22 90.91 90.91 90.91 0.00 0.00 0.00
FECHA 99.74 99.64 99.69 98.93 99.20 99.07 96.65 94.83 95.74 74.95 86.34 80.20
INST 98.96 98.96 98.96 95.73 95.72 95.73 96.08 96.08 96.08 11.11 0.67 1.26
LOC 100 89.45 94.42 94.35 87.88 90.99 92.58 55.55 69.41 0.00 0.00 0.00
NAME 98.99 99.15 99.07 98.97 98.24 98.60 94.78 95.13 94.95 61.62 77.39 68.59
NUM 99.39 99.91 99.65 99.34 98.69 99.01 96.65 97.66 97.15 56.93 68.28 62.05
  99.05 98.96 99.01 98.00 97.34 97.67 94.20 91.10 92.62 62.35 56.11 59.07