In this study, we present a new method for detecting attributes of medical concepts, which uses a sequence labeling approach to recognize attribute entities and classify relations between concepts and attributes simultaneously within one step. A neural architecture combining bidirectional Long Short-Term Memory networks and Conditional Random fields (Bi-LSTMs-CRF) was adopted to detect disorder-modifier pairs in clinical text. Evaluations on the ShARe corpus show that the proposed method achieved higher accuracy and F1 scores than the traditional two-step approaches, indicating its potential to accelerate practical clinical NLP applications.
Keywords: clinical notes; information extraction; natural language processing.