Skip to main content

Table 3 The overall performance of different approaches on the share-disorder dataset in detecting 7 attributes of given disorders: negation (neg), subject (sub), conditional (con), severity (sev), course (cou), uncertainty (unc), body location (bdl). best results are shown in boldface

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

AttributeNEGSUBCONSEVCOUUNCBDL
1.1.1.Baseline
(Bi-LSTM-CRF + SVM)
Acc.0.93230.99290.96690.96550.95760.94450.7524
P0.79310.73740.69900.64210.50680.40910.5887
R0.77680.63480.59870.75680.64370.41720.7516
F0.78490.68220.64490.69480.56710.41310.6602
1.1.1.Baseline
(Bi-LSTM-CRF + Bi-LSTM)
Acc.0.91460.99000.96320.97070.95970.93080.7859
P0.83870.81580.78720.76090.63400.43800.7218
R0.72770.53910.60540.82130.63220.38190.784
F0.77930.64920.68440.79000.63310.40800.7516
1.1.1.Sequence LabelingAcc.0.95420.99370.97180.98170.96970.9550.8695
P0.81420.82220.75830.78120.61500.48540.7887
R0.83100.64350.66820.88590.75290.43930.7991
F0.82250.72200.71040.83020.67700.46120.7939