Chinese medical event detection based on feature extension and document consistency

Extracting valuable medical events from Chinese electronic medical records has important practical significance and application value for electronic medical record text mining, and event detection is a critical step in the event extraction task. Existing research methods for medical event detection in Chinese are mainly based on pattern matching and clustering, and they have two problems $:(1)$ None of them consider the named entities distribution characteristics of medical events. (2) Ignore the distribution of document consistency between medical events in each document. Therefore, this paper proposes an event detection method based on feature extension and document consistency. Firstly, design the medical event representation template and construct the event trigger dictionary according to the ACE standard. Secondly, use semiautomatic corpus labeling method to label entities and events. Then, based on the basic features, according to the distribution characteristics of the entities in the event, different entity information features are selected as extension features. Finally, use the consistency of the distribution of medical documents to improve the final result of event detection. The experimental results show that our method is significantly superior to the baseline.