This is a page that curates AI-related papers published worldwide. All content here is summarized using Google Gemini and operated on a non-profit basis. Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.
MetaSTH-Sleep: Towards Effective Few-Shot Sleep Stage Classification for Health Management with Spatial-Temporal Hypergraph Enhanced Meta-Learning
Created by
Haebom
Author
Jingyu Li, Tiehua Zhang, Jinze Wang, Yi Zhang, Yuhuan Li, Yifan Zhao, Zhishu Shen, Libing Wu, Jiannan Liu
Outline
This paper proposes MetaSTH-Sleep, a novel framework for sleep stage classification. Existing deep learning-based sleep stage classification methods suffer from limitations such as requiring large datasets, poor generalization due to inter-individual differences in biosignals, and ignoring high-dimensional relationships between biosignals. MetaSTH-Sleep is a meta-learning framework utilizing spatiotemporal hypergraphs, employing few-shot learning. It rapidly adapts to new subjects with a small sample size and effectively models the complex spatial interconnections and temporal dynamics of EEG signals. Experimental results demonstrate that MetaSTH-Sleep improves performance across a wide range of subjects.
Takeaways, Limitations
•
Takeaways:
◦
Suggesting the possibility of improving sleep stage classification performance for new subjects even with small amounts of data.
◦
Effectively modeling complex interrelationships in EEG signals using space-time hypergraphs.
◦
Provides useful insights that can assist clinicians in their sleep stage annotation work.
•
Limitations:
◦
Further verification of the generalization performance of the proposed method is needed.
◦
Further research is needed on ways to integrate various biosignals (other than EEG).
◦
Further research is needed to determine its applicability in real clinical settings.
