Skip to main content

Table 4 The overall performance of different approaches on the i2b2-medication dataset in detecting 5 attributes of given medications: dosage (dos), mode (mod), frequency (fre), duration (dur), reason (rea). best results are shown in boldface

From: Applying a deep learning-based sequence labeling approach to detect attributes of medical concepts in clinical text

AttributeDOSMODFREDURREA
Baseline
(Bi-LSTM-CRF + SVM)
Acc.0.92010.95840.93530.97830.9473
P0.87940.91100.87620.59450.5373
R0.92920.95970.93900.66800.6704
F0.90360.93470.90650.62910.5965
Baseline
(Bi-LSTM-CRF + Bi-LSTM)
Acc.0.92500.95590.93020.96800.9269
P0.93050.93720.91980.61680.5984
R0.94340.96580.93990.65250.5717
F0.93690.95130.92980.63410.5848
Sequence LabelingAcc.0.95730.98070.95560.98020.9589
P0.97280.97730.95030.77850.7409
R0.91590.95280.90780.44790.4953
F0.94350.96490.92860.56860.5938
Usyd [14]P0.91890.90730.91420.56040.6687
R0.86780.89150.87950.37090.3319
F0.89260.89940.89650.44640.4436