DI++: A deep learning system for patient condition identification in clinical notes
作者：
Highlights：
• Addressed a real-life problem: patient conditions are identified from clinic charts by resource-intensive manual efforts.
• Proposed a two-step workflow for patient condition identification: disease mention extraction and then classification.
• Designed, developed and deployed a system, DI++, to identify patient conditions from clinical charts.
• Proposed an advanced deep learning model for disease mention classification.
• Extensive evaluation demonstrates the superior performance of DI++ over state-of-the-art approaches.
摘要
•Addressed a real-life problem: patient conditions are identified from clinic charts by resource-intensive manual efforts.•Proposed a two-step workflow for patient condition identification: disease mention extraction and then classification.•Designed, developed and deployed a system, DI++, to identify patient conditions from clinical charts.•Proposed an advanced deep learning model for disease mention classification.•Extensive evaluation demonstrates the superior performance of DI++ over state-of-the-art approaches.
论文关键词：Natural language processing (NLP),Concept extraction,Disease mention extraction,Patient condition classification,Clinical notes,Deep neural network,Deep learning
论文评审过程：Received 23 June 2020, Revised 5 November 2020, Accepted 11 November 2021, Available online 2 December 2021, Version of Record 21 December 2021.
论文官网地址：https://doi.org/10.1016/j.artmed.2021.102224
