脑电图
萧条(经济学)
心理学
计算机科学
人工智能
医学
认知心理学
健康
特征(语言学)
语音识别
自然语言处理
作者
Xiaolong Su,Bin Hu,Zhi‐Hong Guan
出处
期刊:
[Institute of Electrical and Electronics Engineers]
日期:2026-05-15
卷期号:: 2323-2328
标识
DOI:10.1109/ccdc69976.2026.11560166
摘要
In the field of depression diagnosis, using physiological and behavioral signals to identify an individual's depressive state has become an important research direction. Although deep learning methods have made certain progress in electroencephalogram (EEG) depression recognition, they still face many challenges in generating individualized diagnostic information for clinical reference, including significant differences in data distribution among different EEG datasets, insufficient generalization ability of models across datasets, and insufficiently systematic diagnostic logic. To this end, this paper proposes an EEG - text fusion framework based on large language models (LLM) to assist in the clinical assessment of depression. This paper designed the Time-Channel Reconstruction Network (TCRNet) to reconstruct and map the EEG signals of different datasets in terms of Time and Channel dimensions, thereby achieving the alignment representation of multi-source EEG data. By integrating the language understanding and semantic analysis capabilities of LLM, the model can comprehensively process EEG features with patients' self-reports and psychological assessment texts to support diagnostic inferences that are more in line with clinical logic. Meanwhile, the stability of the judgment results has been further enhanced through the majority voting mechanism. The experimental results show that this method achieves a classification accuracy rate of$\text{8 5. 5 2 \%}$on the public EEG depression dataset, indicating that it has certain application value in assisting clinical depression assessment.
科研通智能强力驱动
Strongly Powered by AbleSci AI