Related Experiment Videos
Mmf-re: psychological medicine entity relation extraction model based on multi-level feature enhancement
Zixuan Liu1, Xiaohui Yang1, Zhuo Chang1
1School of Cyber Security and Computer, Hebei University, Baoding, 071000 China.
Health Information Science and Systems
|August 3, 2026
Summary
A new deep learning model, MMF-RE, enhances medical text parsing for psychological medicine by improving feature extraction. This model achieves high accuracy in extracting complex entity relationships, aiding in psychological data analysis.
Area of Science:
- Artificial Intelligence
- Natural Language Processing
- Medical Informatics
Background:
- Deep learning in medical text parsing is advancing.
- Extracting complex entity relationships from medical texts is crucial.
- Existing models struggle with lengthy, complex psychological texts.
Purpose of the Study:
- To propose a novel deep learning model, MMF-RE, for psychological medicine entity relation extraction.
- To address limitations in feature extraction for complex medical texts.
- To improve the accuracy of relation extraction in psychological medicine.
Main Methods:
- Developed MMF-RE model incorporating MFE-BERT for enhanced semantic representation.
- Utilized a multi-unit gated convolutional network for multi-granularity local feature extraction.
- Applied FNNAttention mechanism to strengthen word-level relationships.
Main Results:
- Achieved an F1 score of 88.49% on a self-built psychological medicine dataset.
- Reached an F1 score of 87.08% on a biomedical public dataset.
- Demonstrated superior performance compared to existing evaluation indicators.
Conclusions:
- The MMF-RE model is effective and rational for psychological medicine entity relation extraction.
- The proposed methods significantly improve semantic representation and feature extraction.
- This study offers enhanced capabilities for psychological medicine data analysis.