| 摘 要: 目前,心脑耦合算法的研究较少,研究方法主要为传统统计方法。提出一种基于改进Transformer模型新算法,通过该模型预测EEG到 ECG的关系,并使用余弦相似度来评估预测准确率,实现耦合度分析。在DREAMER和Haaglanden Medisch Centrum Sleep Staging Database数据集上分别对平静和睡眠状态下的数据进行训练。结果显示,平静状态下δ波与心电信号的耦合度最强(准确率86.12%),睡眠状态下β波与心电信号的耦合度最强(准确率96.70%)。该算法能够深度挖掘信号序列细节,直观展现心脑耦合程度,有助于心脑关系的研究。 |
| 关键词: 心脑耦合分析 Transformer 自注意力机制 心电图 脑电图 |
|
中图分类号: TP391
文献标识码: A
|
| 基金项目: 国家中医药管理局中医药创新团队及人才支持计划项目(ZYYCXTD-D-202208) |
|
| Heart-Brain Coupling Algorithm Based on an Improved Transformer Model |
|
SUN Zhenqi, CHEN Zhaoxue
|
(School of Health Science and Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China)
156500839@qq.com; chenzhaoxue@163.com
|
| Abstract: Current research on hear-t brain coupling algorithms is limited, primarily relying on traditional statistical methods. This paper proposes a novel algorithm based on an improved Transformer model, which predicts the
relationship from EEG to ECG signals and uses cosine similarity to evaluate prediction accuracy, thereby enabling coupling analysis. The algorithm was trained on datasets from DREAMER and Haaglanden Medisch Centrum Sleep
Staging Database for both calm and sleep states. Results show that during calm states, δ-waves exhibit the strongest coupling with ECG signals ( accuracy: 86. 12% ), while in sleep states, β-waves demonstrate the highest coupling
(accuracy: 96.70% ). The algorithm effectively captures fine-grained details in signal sequences, visually quantifies hear-t brain coupling intensity, and contributes to research on hear-t brain interactions. |
| Keywords: hear-t brain coupling analysis Transformer sel-f attention mechanism electrocardiogram (ECG) electroencephalogram (EEG) |
