Drug repositioning is a crucial aspect of biomedical research, and predicting drug-disease associations (DDAs) is a critical step in this process. With the development of deep learning and neural network technologies, Graph Convolutional Networks (GCNs) have achieved significant performances in this research field. Although existing DDAs models have made substantial progress, there is still need for improvement in sufficiently utilizing and integrating information from multiple biological entities. In this study, we propose a Drug-Gene-Protein-Disease Heterogeneous Graph Convolutional Network (DGPDHGCN) model for drug-disease association prediction. First, we construct a heterogeneous network from multiple data sources and establish meta-paths based on the topological information of biological entities. Then, the DGPDHGCN model learns representations of drugs and diseases from similarity and association data of those entities. Finally, we define a score function to quantify the associations between drugs and diseases. Through extensive experiments, we demonstrate that DGPDHGCN outperforms baseline models in DDAs prediction tasks in terms of metrics such as AUPR, F1-score, precision, and recall. The source code and experimental datasets can be found in https://github.com/Saxon0918/DGPDHGCN

Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) is a positive-stranded single-stranded RNA virus with an envelope frequently altered by unstable genetic material, making it extremely difficult for vaccines, drugs, and diagnostics to work. Understanding SARS-CoV-2 infection mechanisms requires studying gene expression changes. Deep learning methods are often considered for large-scale gene expression profiling data. Data feature-oriented analysis, however, neglects the biological process nature of gene expression, making it difficult to describe gene expression behaviors accurately. In this article, we propose a novel scheme for modeling gene expression during SARS-CoV-2 infection as networks (gene expression modes, GEM), to characterize their expression behaviors. On this basis, we investigated the relationships among GEMs to determine SARS-CoV-2's core radiation mode. Our final experiments identified key COVID-19 genes by gene function enrichment, protein interaction, and module mining. Experimental results show that ATG10, ATG14, MAP1LC3B, OPTN, WDR45, and WIPI1 genes contribute to SARS-CoV-2 virus spread by affecting autophagy.

Trajectory data contains rich spatial and temporal information. Turning trajectories into graphs and then analyzing them efficiently in an AI-empowered way is a representative branch of trajectory analysis in IoV and ITS environments, which is of great significance. This research attempts to project trajectories onto road networks to predict traffic conditions. Extracting accurate spatio-temporal dependencies is the key to improving the analysis. However, two problems exist in the current study. The first one is the focus on the network structure while ignoring node features, and the second one is that the structure cannot be fully utilized. In addition, the static spatial structure may not accurately reflect the dynamic real spatial dependency. In response to these problems, a novel Spatio-Temporal Graph Convolutional Networks via View Fusion for Trajectory Data Analytics (STFGCN) model is designed. It contains two independent views: the structural view and feature view. The view fusion layer is further designed. It includes an extended graph convolutional module and a causal dilated module. The extended graph convolutional module fully extracts dynamic spatial dependencies, while the causal dilated module captures time tendencies. Stacked view fusion layers and a view fusion module perform fusion operations based on the advantages of the two views, efficiently integrating information from both. Several experiments are performed on two real-world trajectory datasets. The results show that a better prediction performance is obtained, especially on the long-range time prediction task.

Current computer-aided diagnosis system with deep learning method plays an important role in the field of medical imaging. The collaborative diagnosis of diseases by multiple medical institutions has become a popular trend. However, large scale annotations put heavy burdens on medical experts. Furthermore, the centralized learning system has defects in privacy protection and model generalization. To meet these challenges, we propose two federated active learning methods for multicenter collaborative diagnosis of diseases: the Labeling Efficient Federated Active Learning (LEFAL) and the Training Efficient Federated Active Learning (TEFAL). The proposed LEFAL applies a task-agnostic hybrid sampling strategy considering data uncertainty and diversity simultaneously to improve data efficiency. The proposed TEFAL evaluates the client informativeness with a discriminator to improve client efficiency. On the Hyper-Kvasir dataset for gastrointestinal disease diagnosis, with only 65% of labeled data, the LEFAL achieves 95% performance on the segmentation task with whole labeled data. Moreover, on the CC-CCII dataset for COVID-19 diagnosis, with only 50 iterations, the accuracy and F1-score of TEFAL are 0.90 and 0.95, respectively on the classification task. Extensive experimental results demonstrate that the proposed federated active learning methods outperform state-of-the-art methods on segmentation and classification tasks for multicenter collaborative disease diagnosis.

The structure of networks can be efficiently represented using motifs, which are those subgraphs that recur most frequently. One route to understanding the motif structure of a network is to study the distribution of subgraphs using statistical mechanics. In this article, we address the use of motifs as network primitives using the cluster expansion from statistical physics. By mapping the network motifs to clusters in the gas model, we derive the partition function for a network, and this allows us to calculate global thermodynamic quantities, such as energy and entropy. We present analytical expressions for the number of certain types of motifs, and compute their associated entropy. We conduct numerical experiments for synthetic and real-world data sets and evaluate the qualitative and quantitative characterizations of the motif entropy derived from the partition function. We find that the motif entropy for real-world networks, such as financial stock market networks, is sensitive to the variance in network structure. This is in line with recent evidence that network motifs can be regarded as basic elements with well-defined information-processing functions.

Predicting Drug-Drug Interactions (DDIs) is usually a time-consuming and labour-intensive task. The undetected adverse interactions between drugs are a common cause of medical injuries. Recently, with the assistance of deep learning algorithms, the accuracy of prediction in DDIs significantly improved. However, previous methods do not take well into account both adapting to datasets and extracting neighbourhood information on the graph-structured data. In this study, we propose a new framework to fill this gap, named Interaction Prediction Graph Attention Network(IPGAT). This framework consists of two modules, i.e., the embedding module and the prediction module. Inspired by the Graph Embedding and Graph Attention Networks, the embedding module extracts features from graph-structured data with high-order neighbourhoods. Then, it directly transfers to the prediction module without an intermediate process. Our proposed IPGAT presents advantages compared to existing DDI prediction methods. Experimental results on the public DrugBank dataset reveal that IPGAT significantly outperforms the state-of-the-art methods such as AMF&AMFP, Conv-LSTM, Graph Auto-Encoder, etc. The corresponding results increase 6.9% in AUROC and at least 8.5% in AUPR for the retrospective experiment. Further studies verify the efficacy of multi-layer and multi-head in the model. The codes are available at https://github.com/yytfy/IPGAT.

Complex systems often have a latent graph structure. Studying the underlying graph structure will help us to analyze the mechanisms of complex phenomena. However, it is a challenging problem to learn effective graph structures from the data and apply them to downstream tasks. In this paper, we propose an end-to-end graph learning approach for Alzheimer’s syndrome diagnosis based on functional magnetic resonance imaging (fMRI) data of brain regions, which is completely data-driven. The interactions between time-series of each brain region are represented as graph structures, and a multi-head attention mechanism is used to update the representations of the nodes. Then, the graph structures are obtained from the feature sampling of the edges. Finally, the learned graph structure is combined with the left-out time-series data features and the node prior to completing the classification task of the brain network. In comparison with the latest research methods, our approach achieves higher classification accuracy.

The measurement of information on the connections between pairs of nodes is an important part of studying network structures. Real-world network datasets contain fine features on their edges, which are usually represented as edge weights, to reflect the strength of the connection. However, to remove spurious connections and understand the topological structure, network edges are usually represented as binary states, being either connected or not. This is still a controversial issue about how to infer binary adjacency matrices from edge weight distribution. Usually, it is achieved by simply thresholding to distinguish true links and to obtain a set of sparse connections. Previously, tools developed in statistical mechanics have provided effective ways to find the optimal threshold so as to maintain the statistical properties in the network structure. Thermodynamic analogies together with statistical ensembles have been proved to be useful in analysing edge-weighted networks. To extend this work, in this paper, we use Debye’s solid model to describe the probability distribution of edge weights. This models the distribution of edge weights using the mixed Gamma distribution. We treat the derived edge-weight distribution as the combination of two Gamma functions and then apply the Expectation-Maximization algorithm to estimate the corresponding parameters. This gives the optimal threshold to convert weighted networks to binary connections. Numerical analysis shows that Debye’s solid model provides a new way to describe the edge weight probability. Moreover, there exists a phase transition in the low-temperature region, corresponding to a structural transition caused by applying the threshold. Experimental results on real-world weighted networks reveal an improved threshold performance for inferring edge connections from edge weights.

Text restoration plays a vital role in Chinese automatic speech recognition (ASR), which includes punctuation prediction and error correction. However, there are two inevitable challenges for this task. On the one hand, there are no public dataset and model for Chinese punctuation prediction. On the other hand, current text restoration methods for automatic speech recognition only focus on Chinese error correction instead of combining with Chinese punctuation prediction task. To address these problems, a BERT-based text restoration method called TRCA is proposed for Chinese ASR consisting of a Chinese punctuation prediction model and a Chinese error correction model. Experiments demonstrate that the proposed TRCA method outperforms state-of-the-art methods for both punctuation prediction and error correction tasks, among which the proposed TRCA improves the average accuracy to 98% in Chinese punctuation prediction.

Thermodynamic characterisations or analogies have proved to provide powerful tools for the statistical analysis of network populations or time series, together with the identification of structural anomalies that occur within them. For instance, classical Boltzmann statistics together with the corresponding partition function have been used to apply the tools of statistical physics to the analysis of variations in network structure. However, the physical analogy adopted in this analysis, together with the interpretation of the resulting system of particles is sometimes vague and remains an open question. This, in turn, has implications concerning the definition of quantities such as temperature and energy. In this paper, we take a novel view of the thermal characterisation where we regard the edges in a network as the particles of the thermal system. By considering networks with a fixed number of nodes we obtain a conservation law which applies to the particle occupation configuration. Using this interpretation, we provide a physical meaning for the temperature which is related to the number of network nodes and edges. This provides a fundamental description of a network as a thermal system. If we further interpret the elements of the adjacency matrix as the binary microstates associated with edges, this allows us to further extend the analysis to systems with edge-weights. We thus introduce the concept of the canonical ensemble into the thermal network description and the corresponding partition function and then use this to compute the thermodynamic quantities. Finally, we provide numerical experiments on synthetic and real-world data-sets to evaluate the thermal characterisations for both unweighted and weighted networks.

In this paper, we apply ensemble methods from statistical physics to analyse fMRI brain networks for Alzheimer's patients. By mapping the nodes in a network to virtual particles in a thermal system, the microcanonical ensemble and the canonical ensemble are analogous to two different fMRI network representations. These representations are obtained by selecting a threshold on the BOLD time series correlations between nodes in different ways. The microcanonical ensemble corresponds to a set of networks with a fixed fraction of edges, while the canonical ensemble corresponds to the set networks with edges obtained with a fixed value of the threshold. In the former case, there is zero variance in the number of edges in each network, while in the latter case the set of networks have a variance in the number of edges. Ensemble methods describe the macroscopic properties of a network by considering the underlying microscopic characterisations which are in turn closely related to the degree configuration and network entropy. Our treatment allows us to specify new partition functions for fMRI brain networks, and to explore a phase transition in the degree distribution. The resulting method turns out to be an effective tool to identify the most salient anatomical brain regions in Alzheimer's disease and provides a tool to distinguish groups of patients in different stages of the disease.

Fog computing is a promising solution to enable delay-sensitive applications in the Internet of Things (IoT). In this article, based on the simultaneous wireless information and power transfer (SWIPT) technology, we investigate the joint offloading of tasks and energy (JOTE) in fog-enabled IoT networks. Specifically, the task node is allowed to offload energy and tasks to multiple neighboring helper nodes in a time-division multiple access (TDMA) manner. When there are no task queues in the nodes, the offloading decision for each task is independent. We first find the offloading strategy to minimize the task execution delay as well as the energy consumption for a specific task and then, analyze the condition under which the JOTE is beneficial. We show that it becomes more and more desirable to offload both the tasks and the energy from the task node as the number of helper nodes gets large. When there are task queues in the nodes, the offloading decision for each task becomes temporally correlated. We then characterize the optimal strategies to offload the tasks and energy jointly over multiple time slots. An online offloading policy based on the Lyapunov optimization is then proposed to minimize the time average expected delay while stabilizing the system operation. Comprehensive numerical results corroborate our theoretical results and demonstrate the superior performance of the proposed JOTE algorithms.

In this paper, we develop a variational principle from the von Neumann entropy for directed graph evolution. We minimise the change of entropy over time to investigate how directed networks evolve under the Euler-Lagrange equation. We commence from our recent work in which we show how to compute the approximate von Neumann entropy for a directed graph based on simple in and out degree statistics. To formulate our variational principle we commence by computing the directed graph entropy difference between different time epochs. This is controlled by the ratios of the in-degree and out-degrees at the two nodes forming a directed edge. It also reveals how the entropy change is related to correlations between the changes in-degree ratio and in-degree, and their initial values. We conduct synthetic experiments with three widely studied complex network models, namely Erdos-Renyi random graphs, Watts-Strogatz small-world networks, and Barabasi-Albert scale-free networks, to simulate the in-degree and out-degree distribution. Our model effectively captures the directed structural transitions in the dynamic network models. We also apply the method to the real-world financial networks. These networks reflect stock price correlations on the New York Stock Exchange(NYSE) and can be used to characterise stable and unstable trading periods. Our model not only effectively captures how the directed network structure evolves with time, but also allows us to detect periods of anomalous network behaviour.

In this paper, we use the Maxwell-Boltzmann partition function to compute network entropy. The partition function is used to model the energy level population statistics where the network is in thermodynamic equilibrium with a heat-bath. Here the network Hamiltonian operator defines a set of energy levels occupied by particles in thermal equilibrium. These energy levels are given by the eigenvalues of the normalized Laplacian matrix. In other words, we investigate a thermalised version of the system normally studied in spectral graph theory, where the thermalisation accounts for noise in the system. We provide a systematic study of the entropy resulting from this characterization. Compared to previous work based on using von Neumann network entropy, this thermodynamic quantity is effective in characterizing changes of network structure and distinguishing different types of network models (e.g. Erdős-Rényi random graphs, small world networks, and scale free networks). Numerical experiments on real world data-sets are presented to evaluate the qualitative and quantitative differences in performance.