Spatiotemporal Mechanisms Research Articles

A knowledge embedded multimodal pseudo-siamese model for atrial fibrillation detection

Atrial fibrillation (AF) is a common arrhythmia disease with a higher incidence rate. The diagnosis of AF is time-consuming. Although many ECG classification models have been proposed to assist in AF detection, they are prone to misclassifying indistinguishable noise signals, and the context information of long-term signals is also ignored, which impacts the performance of AF detection. Considering all the above problems, we propose a knowledge embedded multimodal pseudo-siamese model. The proposed model comprises a temporal-spatial pseudo-siamese network (TSPS-Net) and a knowledge embedded noise filter module. Firstly, a parallel siamese network architecture is proposed in TSPS-Net to process the multimodal representations. Secondly, a spatiotemporal collaborative fusion mechanism (STCFM) is proposed to fuse multimodal features. Finally, medical knowledge is introduced to design manual features, which are used to distinguish noise and fuse with deep features of ECG to obtain the accurate final result. The model’s performance is verified on the CinC 2017 dataset and the MIT-BIH AF dataset. Experimental results showed that the average accuracy achieved 82.17 and 99.11. The F1 score of our proposed model on the CinC 2017 dataset and MIT-BIH dataset was 0.787 and 0.970, respectively.

Industrial multivariate time-series data anomaly detection incorporating attention mechanisms and adversarial training

ABSTRACT To address the challenges faced in industrial anomaly detection, including data sample imbalance, lack of anomaly labels, and complex spatiotemporal relationships in high-dimensional data, this paper proposes a novel multi-modal time-series anomaly detection model that combines attention mechanisms and adversarial training. In this model, the first step involves utilizing graph attention mechanisms to extract sequence correlation features from multi-modal time-series data, which are then summed with the original data to form a dual-feature-based data representation. Subsequently, a self-supervised learning approach is employed to input this data representation into a variational autoencoder’s encoding-decoding network for reconstruction. Anomaly detection is performed by analyzing the error between the input and reconstructed data. The model also employs spatiotemporal attention mechanisms and adversarial training during reconstruction to enhance feature extraction and model generalization. By comparing our proposed model to five commonly used baseline models, we demonstrate its effectiveness in detecting anomalies in scenarios involving high-dimensional data and imbalanced abnormal samples, demonstrating superior anomaly detection performance, as well as excellent performance on real industrial production and processing datasets.

Cotton2035: From Genomics Research to Optimized Breeding.

Cotton is the world's most important natural fiber crop and serves as an ideal model for studying plant genome evolution, cell differentiation, elongation, and cell wall biosynthesis. The first draft of the cotton genome for Gossypium raimondii, completed in 2012, marked the beginning of global efforts in cotton genomics. Over the past decade, the cotton research community has continued to assemble and refine genomes for both wild and cultivated Gossypium species. With the accumulation of de novo genome assemblies and resequencing data across cotton populations, significant progress has been made in uncovering the genetic basis of key agronomic traits. Achieving the goal of cotton genomics-to-breeding (G2B) will require a deeper understanding of the spatiotemporal regulatory mechanisms involved in genome information storage and expression. We advocate for a cotton ENCODE project to systematically decode the functional elements and regulatory networks within the cotton genome. Technological advances, particularly in single-cell sequencing and high-resolution spatiotemporal omics, will be essential in elucidating these regulatory mechanisms. By integrating multi-omics data, genome editing tools, and artificial intelligence, these efforts will empower the genomics-driven strategies needed for future cotton G2B breeding.

The Combined Effects of Topography and Climate Factors Dominate the Spatiotemporal Evolution of the Ecological Environment in the Yangtze River Economic Belt

ABSTRACTThe Yangtze River Economic Belt (YREB) is a key eco‐environmental protection barrier and economic engine in China. Understanding the characteristics of its eco‐environmental spatiotemporal evolution and mechanisms driving these changes is important for revealing shifts in ecosystem structure and function. However, the governing elements of the ecological environment and how these components interact remain unclear due to the strong coupling impact of urbanization and climate change. This paper constructed a long‐term series remote sensing based ecological index (RSEI) to evaluate the characteristics of eco‐environmental variation of YREB during 2000–2024. Subsequently, the fundamental reasons of the eco‐environmental evolution were quantitatively analyzed using optimal parameters‐based geographical detector (OPGD). Results indicated: (1) The average RSEI of YREB was 0.68, indicating a rather high overall eco‐environmental quality (EEQ), with 73% of the region having RSEI values more than 0.60. The regions with excellent conditions were concentrated in Zhejiang, Jiangxi, Guizhou, and Hunan, while the poorer areas were in northwestern Yunnan, southwestern Sichuan, and western Hubei. (2) The EEQ of YREB mainly showed a degradation trend, with degraded areas exceeding 80% over the past 25 years. In the future, most areas might face higher risks of ecological degradation. (3) Digital elevation model (DEM), temperature (Tem), and population (POP) were key driving factors of the ecological environment. There were significant interactions between certain factors, with the interaction between DEM and potential evapotranspiration (Pet) explaining 62.2% of the ecological quality distribution. The findings assisted the creation of an ecological civilization by offering policy proposals for managing and conserving ecological environments.

Deacetylated SNAP47 recruits HOPS to facilitate autophagosome-lysosome fusion independent of STX17

Autophagy, a conserved catabolic process implicated in a diverse array of human diseases, requires efficient fusion between autophagosomes and lysosomes to function effectively. Recently, SNAP47 has been identified as a key component of the dual-purpose SNARE complex mediating autophagosome-lysosome fusion in both bulk and selective autophagy. However, the spatiotemporal regulatory mechanisms of this SNARE complex remain unknown. In this study, we found that SNAP47 undergoes acetylation followed by deacetylation during bulk autophagy and mitophagy. The acetylation status of SNAP47 is regulated by the acetyltransferase CBP and the deacetylase HDAC2. Notably, the spatiotemporal regulatory dynamics of SNAP47 acetylation differ between bulk autophagy and mitophagy due to distinct regulation on the activity of acetyltransferase and deacetylase. Acetylated SNAP47 inhibits autophagosome-lysosome fusion by indirectly impeding SNARE complex assembly. Mechanistically, deacetylated SNAP47 recruits HOPS components to autophagic vacuoles independently of STX17 and STX17-SNAP47 interaction, while acetylated SNAP47 inhibits this recruitment, consequently leading to the failure of SNARE complex assembly. Taken together, our study uncovers a SNAP47 acetylation-dependent regulatory mechanism governing autophagosome-lysosome fusion by modulating the recruitment of HOPS to autophagic vacuoles without involving STX17, SNAP47-STX17 interaction and ternary SNARE complex formation.

Effects of relative phase position of blades on the cascade performance of highly loaded helium compressor

The operational environment of highly loaded helium compressor necessitates characteristics with narrow flow passages and short blade heights. These characteristics result in more prominent blade tip clearance leakage issues and stronger wake effects compared to air compressors. This study aims to clarify the separation flow structure and the spatiotemporal evolution mechanism of the highly loaded helium compressor cascade and to provide technical support for the aerodynamic stability optimization of such compressors. The present study investigated the effect of the relative positions of the upstream blade wake and downstream blade on cascade loss and separation characteristics of the highly loaded helium compressor with numerical and experimental analysis. The results demonstrate that the proportion of secondary flow losses, including tip leakage vortex of the downstream blade, is the lowest when the circumferential distance between the upstream inlet guide vane wake and the downstream rotor leading edge is 0.5 pitch, with an overall loss weight of 10.8% at the outlet section. Moreover, the proportion of secondary flow losses, including tip leakage vortex of the downstream rotor, is the highest when the circumferential distance between the upstream guide vane wake and downstream rotor leading edge (suction surface side) is 0.25 pitch. The overall loss weight at the outlet section for this position is 21.7%, which is twice that when the circumferential distance is 0.5 pitch. In addition, it can be demonstrated that tip leakage vortex plays a dominant role in vortex losses across all relative positions of the blades.

Observation of the spatiotemporal multiscale evolution mechanism for the interaction between nanosecond pulsed laser and fused silica

Observation of the spatiotemporal multiscale evolution mechanism for the interaction between nanosecond pulsed laser and fused silica

LSTM-based estimation of lithium-ion battery SOH using data characteristics and spatio-temporal attention.

As the primary power source for electric vehicles, the accurate estimation of the State of Health (SOH) of lithium-ion batteries is crucial for ensuring the reliable operation of the power system. Long Short-Term Memory (LSTM), a special type of recurrent neural network, achieves sequence information estimation through a gating mechanism. However, traditional LSTM-based SOH estimation methods do not account for the fact that the degradation sequence of battery SOH exhibits trend-like nonlinearity and significant dynamic variations between samples. Therefore, this paper proposes an LSTM-based lithium-ion SOH estimation method incorporating data characteristics and spatio-temporal attention. First, considering the trend-like nonlinearity of the degradation sequence, which is initially gradual and then rapid, input features are filtered and divided into trend and non-trend features. Then, to address the significant dynamic variations between samples, especially for capacity regeneration,a spatio-temporal attention mechanism is designed to extract spatio-temporal features from multidimensional non-trend features. Subsequently, an LSTM model is built with trend features, spatio-temporal features, and actual capacity as inputs to estimate capacity. Finally, the model is trained and tested on different datasets. Experimental results demonstrate that the proposed method outperforms traditional methods in terms of SOH estimation accuracy and robustness.

Characteristics and Impacts of Pollution and Remediation on Riverine Greenhouse Gas Emissions: A Review

Rivers are not only a vital part of the Earth’s water cycle but also sources and sinks for greenhouse gases (GHGs), exerting a significant influence on the global carbon budget. Rapid urbanization and intense human activities lead to water pollution and river habitat degradation, thereby affecting riverine greenhouse gas (GHG) emissions indirectly. Artificial management and restoration measures taken for rivers further increase the uncertainty of GHG emissions from rivers. In the context of carbon neutrality goals, research on GHG emissions from rivers has gradually become a hot topic. However, there is a scarcity of collective and comparative studies on the spatiotemporal patterns and mechanisms of riverine GHG emissions, especially a lack of summaries exploring the impacts of pollution and restoration on GHG emissions from rivers. This work systematically reviews recent studies concerning the emissions of CO2, CH4, and N2O from rivers, with a particular focus on the characteristics and driving factors. Results have shown that riverine GHG emissions exhibit significant spatiotemporal heterogeneity. Besides hydrological factors such as wind speed, flow velocity, rainfall, and water level, large amounts of pollutants entering rivers strongly affect the production and emission of GHGs, since nutrients, organic matter, heavy metals, microplastics, and antibiotics can alter the biogeochemical processes in river ecosystems. Remediation measures can reduce water pollution levels, but some measures may further increase the emission of GHGs from rivers. This work emphasizes the need for conducting in-depth research on the synergies between treating river pollution and reducing riverine GHG emissions. It also proposes to reinforce the monitoring of GHGs and construct emission databases of rivers for sustainable watershed management.

Decoupling characteristics of county-level land use carbon emissions and ecological environment quality at spatiotemporal scales: a case study of Shaanxi Province, China

ABSTRACT A critical approach to achieving dual carbon targets is to maintain strong decoupling (SD) between land use carbon emissions (LUCE) and ecological environment quality (EEQ). However, the spatiotemporal decoupling mechanisms of LUCE and EEQ at the county scale remain understudied. This study aims to explore the decoupling dynamics of LUCE and EEQ across 107 counties in Shaanxi Province from 2000 to 2020. Using the PIE-Engine and Google Earth Engine, we constructed LUCE and Remote Sensing Ecological Index (RSEI) models and applied the Tapio decoupling theory to analyze the decoupling trends. The results reveal that from 2000 to 2020, LUCE in Shaanxi tripled, while EEQ increased by 20.93%, albeit with fluctuations. County-level variations in LUCE and EEQ exhibited pronounced spatiotemporal heterogeneity, with decoupling types undergoing significant transitions in over 95% of counties. In 2005, 28.04% of counties achieved a decoupled state, but this deteriorated sharply by 2020, when nearly half of the counties displayed strong negative decoupling (SND), and no counties maintained SD. These findings suggest that achieving SD remains challenging and requires targeted strategies based on regional decoupling characteristics. This study offers a theoretical and practical framework for understanding county-level LUCE and EEQ decoupling, which is crucial for sustainable development.

Landscape Characteristics Influencing the Spatiotemporal Dynamics of Soundscapes in Urban Forests

The acoustic environment of urban forests is indispensable for urban residents’ nature-based recreation opportunities and experience of green spaces, and the perceptual and physical sound features in time and space serve as determinants during this process. However, their spatiotemporal variation mechanisms and influential landscape characteristics are still underexplored in urban forests. Thus, this study aims to explore the spatiotemporal variability of perceptual and physical sound features and their relationship with landscape characteristics in urban forests. For this purpose, we measured perceptual sound features using the indicators of the sound harmonious degree (SHD) and soundscape pleasantness and eventfulness. The physical acoustic features were determined using sound-level parameters for measuring the sound level intensity (LAeq, L10, L90) and fluctuation (L10–90). Perceptual and physical sound data collection was based on on-site questionnaire surveys and acoustic instrument measurements, respectively. The landscape characteristics were classified using the principal components of four main categories, including the terrain, area proportion of land cover types, distance to land cover types, and landscape patterns. The results showcase that significant spatiotemporal variation was found in most perceptual and physical sound features, whereas soundscape pleasantness and eventfulness did not vary significantly across time. In general, the variabilities of both perceptual and physical sound features were affected more by the types of spatial functions than by diurnal patterns. Human activities that generate sounds (e.g., hawking, playing, and exercise) may be the key drivers for spatiotemporal changes in physical acoustic features. The components of landscape patterns, including landscape structural diversity and shape complexity persistently, affected specific sound features in all periods. However, no landscape component had persistent cross-spatial influences on the sound features. This study offers critical insights into the spatiotemporal patterns of the acoustic environment and its relationship with landscape characteristics in urban forests. The findings underscore the practical importance and implications of integrating acoustic considerations into urban forest management. By providing a scientific foundation, these results can usefully inform dynamic resource management, functional zoning optimization, and sustainable landscape development in urban forests.

Synergistic Matching and Influencing Factors of Grain Production and Cropland Net Primary Productivity in the Black Soil Region of Northeast China

Exploring the spatiotemporal dynamics, spatial mismatch, and complex influencing mechanism of grain production and cropland productivity in the black soil region of northeast China (BSRNC) is essential for the synergistic protection and utilization of black soil cropland and sustainable grain production. The BSRNC has realized cropland expansion and grain production increases in the past decades. This implied a substantial investment has been made in the region’s agriculture. However, at present, knowledge on the spatial mismatch and influencing factors of grain production and cropland productivity is still unclear. This study analyzed the spatial–temporal mismatch characteristics of grain production and cropland net primary productivity (CNPP) using the gravity center model, spatial autocorrelation analysis, and spatial mismatch index (SMI), and identified the spatial heterogeneity and prediction–response relationships of influencing factors based on a geographically and temporally weighted regression (GTWR) model and boosted regression tree (BRT) machine learning algorithm. The findings indicated that grain production and CNPP have been increasing, but the overall spatial pattern of cold hotspots has not changed obviously in the BSRNC from 2000 to 2020. The SMI has shown a decreasing trend, indicating that the synergistic development of grain production and CNPP has been obvious, which plays an important role in sustainable food supply capacity. Agricultural production and the natural environment have always been critical factors influencing the spatial mismatch. Specifically, the marginal impact of fertilizer application has undergone a shift. This study may provide new clues for the formulation of regional strategies for sustainable food supply and black soil cropland system protection.

Assessing the Spatio-Temporal Evolution Mechanisms from Meteorological to Hydrological Drought in a Heavily Human-Influenced River Basin

Assessing the Spatio-Temporal Evolution Mechanisms from Meteorological to Hydrological Drought in a Heavily Human-Influenced River Basin

A novel spatio-temporal attention mechanism model for car-following in autonomous driving

Car following is critical for the overall safety, efficiency, and smooth operation of autonomous vehicles in traffic. However, existing car-following model primarily focuses on local feature extraction, overlooking the spatial–temporal relationships between vehicles and key information within the time series. This limitation negatively impacts prediction accuracy and generalization capabilities. To address these issues, a novel Spatio-Temporal Car-Following model based on Graph Convolution Network and Attention mechanism is proposed to enhance the safety, efficiency, and comfort of autonomous driving systems. Firstly, the spatial–temporal structure of the car-following model is constructed using the GCN network. This approach efficiently captures the topological relationships between vehicles. Secondly, we interleave spatial and temporal blocks to extract feature information from both spatial and temporal dimensions, which allows us to perceive spatial interactions and temporal motion patterns of the car. Additionally, a self-attention module assigns different attention weights to the various neighbors of a node in the car-following model, which can facilitate the capture of diverse aspects of node relationships and enhance expressive power. Finally, a Multi-Layer Perceptron (MLP) layer predicts the future behaviors of following vehicles. The proposed car-following model (CF) was trained and evaluated using five real-world datasets: HighD, SPMD, Waymo, NGSIM, and Lyft. The results indicate that our approach surpasses current models in terms of Mean Squared Error (MSE) for spacing, while also maintaining a zero collision rate across every dataset. These findings indicate that the Spatio-Temporal Attention Model (GSTAM-CF) proposed in this paper enhances safety, efficiency, and operational comfort compared to existing models.

Action recognition in rehabilitation: combining 3D convolution and LSTM with spatiotemporal attention.

This study addresses the limitations of traditional sports rehabilitation, emphasizing the need for improved accuracy and response speed in real-time action detection and recognition in complex rehabilitation scenarios. We propose the STA-C3DL model, a deep learning framework that integrates 3D Convolutional Neural Networks (C3D), Long Short-Term Memory (LSTM) networks, and spatiotemporal attention mechanisms to capture nuanced action dynamics more precisely. Experimental results on multiple datasets, including NTU RGB + D, Smarthome Rehabilitation, UCF101, and HMDB51, show that the STA-C3DL model significantly outperforms existing methods, achieving up to 96.42% accuracy and an F1 score of 95.83% on UCF101, with robust performance across other datasets. The model demonstrates particular strength in handling real-time feedback requirements, highlighting its practical application in enhancing rehabilitation processes. This work provides a powerful, accurate tool for action recognition, advancing the application of deep learning in rehabilitation therapy and offering valuable support to therapists and researchers. Future research will focus on expanding the model's adaptability to unconventional and extreme actions, as well as its integration into a wider range of rehabilitation settings to further support individualized patient recovery.

Disentangled Dynamic Graph Attention Network for Out-of-Distribution Sequential Recommendation

Sequential recommendation, leveraging user-item interaction histories to provide personalized and timely suggestions, has drawn significant research interest recently. With the power of exploiting spatio-temporal dynamics, Dynamic Graph Neural Networks (DyGNNs) show great potential in sequential recommendation by modeling the dynamic relationship between users and items. However, spatio-temporal distribution shifts naturally exist in out-of-distribution sequential recommendation, where both user-item relationships and temporal sequences demonstrate pattern shifts. The out-of-distribution scenarios may lead to the failure of existing DyGNNs in handling spatio-temporal distribution shifts in sequential recommendation, given that the patterns they exploit tend to be variant w.r.t labels under distribution shifts. In this article, we propose Disentangled Intervention-based Dynamic graph Attention networks with Invariance Promotion ( I-DIDA ) to handle spatio-temporal distribution shifts in sequential recommendation by discovering and utilizing invariant patterns , i.e., structures and features whose predictive abilities are stable across distribution shifts. Specifically, we first propose a disentangled spatio-temporal attention network to capture the variant and invariant patterns. By utilizing the disentangled patterns, we design a spatio-temporal intervention mechanism to create multiple interventional distributions and an environment inference module to infer the latent spatio-temporal environments, and minimize the invariance loss to leverage the invariant patterns with stable predictive abilities under distribution shifts. Extensive experiments demonstrate the superiority of our method over state-of-the-art sequential recommendation baselines under distribution shifts.

Spatio-temporal influencing effects and mechanisms of the digital economy on eco-urbanization in the Yangtze River Delta region

Spatio-temporal influencing effects and mechanisms of the digital economy on eco-urbanization in the Yangtze River Delta region

Spatio-temporal attention-based hidden physics-informed neural network for remaining useful life prediction

Predicting the Remaining Useful Life (RUL) is essential in Prognostic Health Management (PHM) for industrial systems. Although deep learning approaches have achieved considerable success in predicting RUL, challenges such as low prediction accuracy and interpretability pose significant challenges, hindering their practical implementation. In this work, we introduce a Spatio-temporal Attention-based Hidden Physics-informed Neural Network (STA-HPINN) for RUL prediction, which can utilize the associated physics of the system degradation. The spatio-temporal attention mechanism can extract important features from the input data. With the self-attention mechanism on both the sensor dimension and time step dimension, the proposed model can effectively extract degradation information. The hidden physics-informed neural network is utilized to capture the physics mechanisms that govern the evolution of RUL. With the constraint of physics, the model can achieve higher accuracy and reasonable predictions. The approach is validated on a benchmark dataset, demonstrating exceptional performance when compared to cutting-edge methods, especially in the case of complex conditions.

Action Mechanisms of Exosomes Derived from GD3/GD2-Positive Glioma Cells in the Regulation of Phenotypes and Intracellular Signaling: Roles of Integrins.

Extracellular vesicles (EVs) play important roles in intercellular communication in various biological events. In particular, EVs released from cancer cells have attracted special attention. Although it has been reported that cancer-associated glycosphingolipids play important roles in the enhancement of malignant properties of cancer cells, the presence, behavior, and roles of glycosphingolipids in EVs have not been elucidated. Recently, we reported crucial roles of EVs expressing gangliosides, GD2, and/or GD3 in the enhancement of cancer properties in malignant melanomas and gliomas. However, how EVs containing cancer-associated glycosphingolipids play their roles has not been reported to date. Here, we studied spatio-temporal mechanisms for GD3/GD2-containing EVs released from gliomas in the actions toward target cells. Proteome analyses of EVs with/without GD3/GD2 revealed an equally high concentration of integrin isoforms in both GD3/GD2+ and GD3/GD2- EVs. PKH26-labeled EVs attached, invaded, and distributed to/in the target cells within 1 h. GD3/GD2 formed molecular complexes with integrins on EVs as elucidated by immunoprecipitation/immunoblotting and immunocytostaining. The addition of antibodies reactive with GD3, GD2, or integrins resulted in the suppression of the enhancing effects of EVs in the cell adhesion assay. The addition of GD3/GD2 + EVs to GD3/GD2- cells clearly increased the phosphorylation levels of the PDGF receptor, FAK, and Erk1/2 in immunoblotting, suggesting GD3/GD2+ EVs activate the signaling pathway in the target cells within 15 min after addition. Anti-ganglioside antibodies clearly blocked signaling with EVs. In conclusion, EVs released from GD3/GD2-expressing glioma cells enhance cancer phenotypes and malignant signals via the cluster formation of integrins and GD3/GD2 on EVs, leading to the regulation of the cancer microenvironment.

Integrating Historical Learning and Multi-View Attention with Hierarchical Feature Fusion for Robotic Manipulation.

Humans typically make decisions based on past experiences and observations, while in the field of robotic manipulation, the robot's action prediction often relies solely on current observations, which tends to make robots overlook environmental changes or become ineffective when current observations are suboptimal. To address this pivotal challenge in robotics, inspired by human cognitive processes, we propose our method which integrates historical learning and multi-view attention to improve the performance of robotic manipulation. Based on a spatio-temporal attention mechanism, our method not only combines observations from current and past steps but also integrates historical actions to better perceive changes in robots' behaviours and their impacts on the environment. We also employ a mutual information-based multi-view attention module to automatically focus on valuable perspectives, thereby incorporating more effective information for decision-making. Furthermore, inspired by human visual system which processes both global context and local texture details, we have devised a method that merges semantic and texture features, aiding robots in understanding the task and enhancing their capability to handle fine-grained tasks. Extensive experiments in RLBench and real-world scenarios demonstrate that our method effectively handles various tasks and exhibits notable robustness and adaptability.