Early prediction of sepsis is critical for reducing patient mortality. However, real-world electronic health records (EHRs) often contain irregularly sampled multivariate time series with high missingness. Although transformer-based models have shown strong performance in time series analysis, their global attention mechanism struggles to effectively capture mutual influence among multiple vital signs and temporal dynamic features. To address these challenges, we proposed a novel framework with Dual-Axial Warped Networks (DAWN) for early sepsis prediction on irregular time series data. It consists of two key components: a warping module and a dual-axial attention aggregation module. The dual-axial attention mechanism includes two complementary feature views: one view applies axial at-tention along the variable dimension, followed by a deformable neighborhood attention mechanism, which can aggregate shifted receptive fields along the temporal dimension and enhance the sensitivity of local feature variations. The other view performs axial attention along the temporal dimension, then aggregates information along variables to reinforce integrated modeling of temporal dynamics and inter-variable interactions. Both view features are adaptively fused via a gating mechanism to enable personalized representation of patient states. Additionally, the warping module learns unified warped representations of irreg-ular time series on multiple scales to mitigate high missingness. The experiments were carried out on two public datasets. The proposed model outperforms existing methods with AUPRC scores of 60.8 and 73.4 on two datasets, respectively, which demonstrates the effectiveness of our framework.

Recent studies increasingly explore the use of Electronic Health Records (EHR)-including diagnostic, surgical, and historical visit data-for medication recommendation. An effective EHR representation must integrate both the temporal relationships across visits and medical events, and their intrinsic clinical features. Existing methods fall into three categories: instance-based, sequential, and graphical representations. Instance-based approaches rely only on current visit data, ignoring patient history and leading to limited accuracy. Sequential methods focus solely on visit timelines, while graphical ones capture structural relationships among medical events but overlook temporal dynamics. To combine the strengths of both, we propose STRMeR, a novel model that enhances medication recommendation by mining complex structures and temporal patterns from EHR. It dynamically tracks a patient's health state by propagating structural information from medical events to visits and incorporating temporal visit embeddings. We also introduce an Adaptive Patch Mechanism to identify local time-related patterns in a data-driven way, and a Multi-Scale Convolution Mechanism to capture short-, medium-, and longterm temporal dependencies. Evaluated on the MIMIC-III and MIMIC-IV benchmarks, STRMeR outperforms current state-of-the-art methods.

Predicting the binding affinity of drug target is essential to reduce drug development costs and cycles. Recently, several deep learning-based methods have been proposed to utilize the structural or sequential information of drugs and targets to predict the drug-target binding affinity (DTA). However, methods that rely solely on sequence features do not consider hydrogen atom data, which may result in information loss. Graph-based methods may contain information that is not directly related to the prediction process. Additionally, the lack of structured division can limit the representation of characteristics. To address these issues, we propose a multimodal DTA prediction model using graph local substructures, called MLSDTA. This model comprehensively integrates the graph and sequence modal information from drugs and targets, achieving multimodal fusion through a cross-attention approach for multimodal features. Additionally, adaptive structure aware pooling is applied to generate graphs containing local substructural information. The model also utilizes the DropNode strategy to enhance the distinctions between different molecules. Experiments on two benchmark datasets have shown that MLSDTA outperforms current state-of-the-art models, demonstrating the feasibility of MLSDTA.

Single-cell type annotation helps to identify specific subpopulations of cells associated with a particular disease, thus providing therapeutic targets for precision medicine. Previous tasks on single-cell type annotation have been performed on the basis of RNA-seq sequencing data, and gradually some studies are now being conducted on scATAC-seq sequencing data as well. Although the researchs in this area has achieved considerable results, it is still constrained by the high dimensionality, high noise, and uneven sample distribution inherent in scATAC-seq datasets. In this paper, we propose a graph convolution model that incorporates ensemble learning. The utilization of Pseudo Labels for cell graph construction significantly mitigates the excessively high anisotropy index in the graph caused by noise, presenting a novel approach to enhancing graph quality. Furthermore, our method introduces a second innovation by effectively addressing the challenge of high-dimensional features in ATAC data through guided feature selection facilitated by ensemble learning, ultimately improving the overall performance and accuracy of the analysis. The model effectively reduces the impact of the previously mentioned problems on the scATAC-seq dataset, and improves the predictive performance of the model. The single-cell type annotation results exceed 90% of the ACC in different datasets and the Macro-average ACC exceeds that of other models.

With the prevalence of cloud storage, people prefer to outsource their data to the cloud for flexibility and reliability. Undoubtedly, there are lots of redundancy among these data. However, high-end storage with deduplication costs heavy computation and increases the data management complexity. Potential customers need the redundancy proportion information of their outsourced data to decide whether high-end storage with deduplication is worthwhile. Thus, many researchers have previously attempted to predict the redundant ratio. However, existing mechanisms ignore the redundancy proportion among similar chunks containing many duplicate data. Although resemblance detection, detecting the duplicate parts among similar data, has become a hot issue, it is hardly applied to the conventional deduplication ratio estimation because of unacceptable calculation cost. Therefore, we analyze the limitations and challenges of deduplication ratio prediction in prediction scope and response time and further propose a novel prediction scheme. By leveraging the context-aware resemblance detection, and confidence interval theory, our method can achieve faster estimation speed with higher accuracy in deduplication ratio compared with the state-of-the-art work. Finally, the results show that our method can efficiently and effectively estimate the proportion of duplicate chunks and redundant data among similar chunks by conducting experiments on real workloads.

BACKGROUND: Representation learning provides new and powerful graph analytical approaches and tools for the highly valued data science challenge of mining knowledge graphs. Since previous graph analytical methods have mostly focused on homogeneous graphs, an important current challenge is extending this methodology for richly heterogeneous graphs and knowledge domains. The biomedical sciences are such a domain, reflecting the complexity of biology, with entities such as genes, proteins, drugs, diseases, and phenotypes, and relationships such as gene co-expression, biochemical regulation, and biomolecular inhibition or activation. Therefore, the semantics of edges and nodes are critical for representation learning and knowledge discovery in real world biomedical problems. RESULTS: In this paper, we propose the edge2vec model, which represents graphs considering edge semantics. An edge-type transition matrix is trained by an Expectation-Maximization approach, and a stochastic gradient descent model is employed to learn node embedding on a heterogeneous graph via the trained transition matrix. edge2vec is validated on three biomedical domain tasks: biomedical entity classification, compound-gene bioactivity prediction, and biomedical information retrieval. Results show that by considering edge-types into node embedding learning in heterogeneous graphs, edge2vec significantly outperforms state-of-the-art models on all three tasks. CONCLUSIONS: We propose this method for its added value relative to existing graph analytical methodology, and in the real world context of biomedical knowledge discovery applicability.

During a search task, a user’s search intention is possible inaccurate. Even with clear information need, it is probable that the search query cannot precisely describe the user’s need. And besides, the user is utterly impossible browse all the returned results. Thus, a selected and valuable returned search list is quite important for a search system. Actually, there are lots of reliable and highly relevant personal documents existing in a user’s personal computer. Based on the desktop documents, it is relevantly easy to understand the user’s current knowledge level about the present search subject, which is useful to predict a user’s need. An approach was proposed to exploit the potential of desktop context to refine the search returned list. Firstly, to attain a comprehensive long-term user model, the operational history and a series of time-related information were analyzed to achieve the attention degree that a user paid to a document. And the keywords and user tags were focused on to understand the content. Secondly, working scenario was regarded as the most valuable information to construct a short-term user model, which directly suggested what exactly a user was working on. Experiment results showed that desktop context could effectively help refine the search returned results, and only the effectively combination of the long-term user model and the short-term user model could offer more relevant items to satisfy the user.

With the development of e-science, more and more attention is drawn to the visualization and interoperation of scientific data. Integrating Data Grid technology and Geographical Information System (GIS) offers a well solution to implement e-science applications. Utilizing Data Grid technology could take advantages to share heterogeneous and distributed data sources. GIS provides visualization environment for spatial data and non-spatial data. In this paper we brought forward spatial information grid (SIG) architecture for gas network safety evaluation application. After discussing the key technologies of gas network SIG, such as spatial metadata modeling and spatial metadata searching, one prototype system was designed and then implemented. Through the SIG system, gas network safety evaluation results can be shown to users in an intuitive and convenient way.