Chinese medical named entity recognition utilizing entity association and gate context awareness

Area of Science:

• Natural Language Processing
• Medical Informatics
• Artificial Intelligence

Background:

• Accurate medical named entity recognition (NER) is vital for deep learning applications in healthcare.
• Current NER methods struggle with contextual awareness and inter-entity relationships.
• Efficient information extraction from medical texts is crucial for improving healthcare services and clinical decisions.

Purpose of the Study:

• To develop an advanced deep learning model for Chinese medical NER.
• To address limitations in contextual understanding and entity interaction in existing NER approaches.
• To enhance the accuracy and efficiency of processing medical literature.

Main Methods:

• Utilized the Chinese pre-trained RoBERTa-wwm-ext model for contextual feature extraction.
• Employed recurrent neural networks with multi-head attention for parallel processing and capturing inter-entity dependencies.
• Integrated conditional random fields with cross-entropy loss for improved accuracy and label sequence consistency.

Main Results:

• The proposed model achieved an F1 score of 91.90% on the MCSCSet dataset.
• The model attained an F1 score of 64.36% on the CMeEE dataset.
• Outperformed existing related models in Chinese medical NER tasks.

Conclusions:

• The developed model demonstrates significant efficacy in recognizing named entities within Chinese medical texts.
• The approach effectively captures contextual information and inter-entity dependencies.
• This advancement holds promise for enhancing medical information processing and clinical support systems.