Multi-View Spatial-Temporal Graph Convolutional Networks With Domain Generalization for Sleep Stage Classification

Jia, Zhensheng; Lin, Youfang; Wang, Jing; Ning, Xiaojun; He, Yuanlai; Zhou, Ronghao; Zhou, Yuhan; Lehman, Li-wei H.

doi:10.1109/tnsre.2021.3110665

Cited by 140 publications

(72 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The study of spatial-temporal data has attracted increasing attention in recent years, especially after the emergence of many advanced deep learning models. It is pointed out that spatial-temporal data could benefit applications in many disciplines such as emotion detection, traffic flow prediction, and sleep stage/quality classification [6]. However, It has been a long-standing challenge to better uncover and understand the hidden attributes in these numeric data.…”

Section: Related Work a Deep Learning For Spatial-temporal Datamentioning

confidence: 99%

“…{X (1) , ..., X (t)} = transform ({S (1) , ..., S (t)} ; θ S ) (6) where θ S is the trainable parameters in the feature transformation net, and X = {X (1) , X (2) , ..., X (t)} ∈ R T ×N ×d is the high-quality output, i.e. the generated initial node features.…”

Section: A Feature Transformation Netmentioning

confidence: 99%

“…Additionally, one common issue is that both model types require the grid-like data input, which is considered intractable when decoding the topological information from the spatial dimension. It is argued that the non-Euclidean graph data structure shows great potential in terms of representing inter-connectivity and dynamic trends in both spatial and temporal aspects [6]. Recently, modelling the spatial-temporal data into graph representation has risen to the spotlight owing to the development of graph neural network (GNN), reporting some encouraging improvements on both classification and regression tasks compared with CNN and RNN based models [7], [8].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

FedRel: An Adaptive Federated Relevance Framework for Spatial Temporal Graph Learning

Zhang¹,

Liu²,

Shen³

et al. 2022

Preprint

View full text Add to dashboard Cite

Spatial-temporal data contains rich information and has been widely studied in recent years due to the rapid development of relevant applications in many fields. For instance, medical institutions often use electrodes attached to different parts of a patient to analyse the electorencephal data rich with spatial and temporal features for health assessment and disease diagnosis. Existing research has mainly used deep learning techniques such as convolutional neural network (CNN) or recurrent neural network (RNN) to extract hidden spatialtemporal features. Yet, it is challenging to incorporate both inter-dependencies spatial information and dynamic temporal changes simultaneously. In reality, for a model that leverages these spatial-temporal features to fulfil complex prediction tasks, it often requires a colossal amount of training data in order to obtain satisfactory model performance. Considering the abovementioned challenges, we propose an adaptive federated relevance framework, namely FedRel, for spatial-temporal graph learning in this paper. After transforming the raw spatialtemporal data into high quality features, the core Dynamic Inter-Intra Graph (DIIG) module in the framework is able to use these features to generate the spatial-temporal graphs capable of capturing the hidden topological and long-term temporal correlation information in these graphs. To improve the model generalization ability and performance while preserving the local data privacy, we also design a relevance-driven federated learning module in our framework to leverage diverse data distributions from different participants with attentive aggregations of their models. In addition, we conduct extensive experiments on two real-world spatial-temporal datasets. The results demonstrate the effectiveness of our proposed framework in spatial-temporal interpretation, collaborative model training, and divergent data distribution handling in most settings of comparison.

show abstract

Section: Related Work a Deep Learning For Spatial-temporal Datamentioning

confidence: 99%

Section: A Feature Transformation Netmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

FedRel: An Adaptive Federated Relevance Framework for Spatial Temporal Graph Learning

Zhang¹,

Liu²,

Shen³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…However, the dependence on non-adjacent electrodes placed in different brain regions is often overlooked. Since then, Jia et al [ 30 ] propose an multi-view spatial-temporal graph convolutional network (MSTGCN) to extract the most relevant spatial and temporal information with superior performance. They introduce spatiotemporal attention to extract temporal and spatial information, respectively.…”

Section: Introductionmentioning

confidence: 99%

An Attention-Guided Spatiotemporal Graph Convolutional Network for Sleep Stage Classification

Chen

Cheng

2022

Life

View full text Add to dashboard Cite

Sleep staging has been widely used as an approach in sleep diagnoses at sleep clinics. Graph neural network (GNN)-based methods have been extensively applied for automatic sleep stage classifications with significant results. However, the existing GNN-based methods rely on a static adjacency matrix to capture the features of the different electroencephalogram (EEG) channels, which cannot grasp the information of each electrode. Meanwhile, these methods ignore the importance of spatiotemporal relations in classifying sleep stages. In this work, we propose a combination of a dynamic and static spatiotemporal graph convolutional network (ST-GCN) with inter-temporal attention blocks to overcome two shortcomings. The proposed method consists of a GCN with a CNN that takes into account the intra-frame dependency of each electrode in the brain region to extract spatial and temporal features separately. In addition, the attention block was used to capture the long-range dependencies between the different electrodes in the brain region, which helps the model to classify the dynamics of each sleep stage more accurately. In our experiments, we used the sleep-EDF and the subgroup III of the ISRUC-SLEEP dataset to compare with the most current methods. The results show that our method performs better in accuracy from 4.6% to 5.3%, in Kappa from 0.06 to 0.07, and in macro-F score from 4.9% to 5.7%. The proposed method has the potential to be an effective tool for improving sleep disorders.

show abstract

“…Since 1980, exercise has been incorporated into the health management system. After years of development, the United States has established an exercise promotion and health guidance service platform that is led by the government, supplemented by scientific research institutions and sports social organizations, and combines medical and health services with physical fitness services [20][21][22][23][24]. Based on inspecting the American sports promotion health guidance service platform, from the establishment of the "medicine and body integration" linkage management mechanism, to play the role of social organizations such as the Chinese Society of Sports Science, to refine the implementation goals of sports to promote health, and to strengthen compound sports to promote health.…”

Section: Introductionmentioning

confidence: 99%

Management and Analysis of Sports Health Level of the Elderly Based on Deep Learning

Xiao

Li-min

Chang

et al. 2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

With the accelerating rate of population aging in China, the health of the elderly has received more and more attention and has become one of the most important issues in the elderly care industry. Because of insufficient research on the personal health of the elderly, the value of medical examination data cannot be fully exploited, many physical indicators have a certain impact on overall health or heart health, and there are few studies on heart health assessment. This paper proposes a deep learning-based elderly management analysis method of human exercise health level, using the exercise health management model to evaluate the heart health level of the elderly. Firstly, the indicators to measure heart health are proposed through traditional expert knowledge and personal health index to analyze heart health. Through dynamic assessment, predict the heart health status at the next time point, analyze possible heart diseases, and provide corresponding methods for the health of the elderly, which helps improve the physical health of the elderly. Quality of life provides assistance to meet the needs of improving the health of older adults.

show abstract

Multi-View Spatial-Temporal Graph Convolutional Networks With Domain Generalization for Sleep Stage Classification

Cited by 140 publications

References 35 publications

FedRel: An Adaptive Federated Relevance Framework for Spatial Temporal Graph Learning

FedRel: An Adaptive Federated Relevance Framework for Spatial Temporal Graph Learning

An Attention-Guided Spatiotemporal Graph Convolutional Network for Sleep Stage Classification

Management and Analysis of Sports Health Level of the Elderly Based on Deep Learning

Contact Info

Product

Resources

About