Attributed Multi-Order Graph Convolutional Network for Heterogeneous Graphs

Heterogeneous graph neural networks play a crucial role in discovering discriminative node embeddings and relations from multi-relational networks. One of the key challenges in heterogeneous graph learning lies in designing learnable meta-paths, which significantly impact the quality of learned embeddings. In this paper, we propose an Attributed Multi-Order Graph Convolutional Network (AMOGCN), which automatically explores meta-paths that involve multi-hop neighbors by aggregating multi-order adjacency matrices. The proposed model first constructs different orders of adjacency matrices from manually designed node connections. Next, AMOGCN fuses these various orders of adjacency matrices to create an intact multi-order adjacency matrix. This process is supervised by the node semantic information, which is extracted from the node homophily evaluated by attributes. Eventually, we employ a one-layer simplifying graph convolutional network with the learned multi-order adjacency matrix, which is equivalent to the cross-hop node information propagation with multi-layer graph neural networks. Substantial experiments reveal that AMOGCN achieves superior semi-supervised classification performance compared with state-of-the-art competitors.

Mesh neural networks for SE(3)-equivariant hemodynamics estimation on the artery wall

Computational fluid dynamics (CFD) is a valuable asset for patient-specific cardiovascular-disease diagnosis and prognosis, but its high computational demands hamper its adoption in practice. Machine-learning methods that estimate blood flow in individual patients could accelerate or replace CFD simulation to overcome these limitations. In this work, we consider the estimation of vector-valued quantities on the wall of three-dimensional geometric artery models. We employ group-equivariant graph convolution in an end-to-end SE(3)-equivariant neural network that operates directly on triangular surface meshes and makes efficient use of training data. We run experiments on a large dataset of synthetic coronary arteries and find that our method estimates directional wall shear stress (WSS) with an approximation error of 7.6% and normalised mean absolute error (NMAE) of 0.4% while up to two orders of magnitude faster than CFD. Furthermore, we show that our method is powerful enough to accurately predict transient, vector-valued WSS over the cardiac cycle while conditioned on a range of different inflow boundary conditions. These results demonstrate the potential of our proposed method as a plugin replacement for CFD in the personalised prediction of hemodynamic vector and scalar fields.

DRTerHGAT: A drug repurposing method based on the ternary heterogeneous graph attention network

Drug repurposing is an effective method to reduce the time and cost of drug development. Computational drug repurposing can quickly screen out the most likely associations from large biological databases to achieve effective drug repurposing. However, building a comprehensive model that integrates drugs, proteins, and diseases for drug repurposing remains challenging. This study proposes a drug repurposing method based on the ternary heterogeneous graph attention network (DRTerHGAT). DRTerHGAT designs a novel protein feature extraction process consisting of a large-scale protein language model and a multi-task autoencoder, so that protein features can be extracted accurately and efficiently from amino acid sequences. The ternary heterogeneous graph of drug-protein-disease comprehensively considering the relationships among the three types of nodes, including three homogeneous and three heterogeneous relationships. Based on the graph and the extracted protein features, the deep features of the drugs and the diseases are extracted by graph convolutional networks (GCN) and heterogeneous graph node attention networks (HGNA). In the experiments, DRTerHGAT is proven superior to existing advanced methods and DRTerHGAT variants. DRTerHGAT's powerful ability for drug repurposing is also demonstrated in Alzheimer's disease.

HR-BGCN : Predicting readmission for heart failure from electronic health records

Heart failure has become a huge public health problem, and failure to accurately predict readmission will further lead to the disease's high cost and high mortality. The construction of readmission prediction model can assist doctors in making decisions to prevent patients from deteriorating and reduce the cost burden. This paper extracts the patient discharge records from the MIMIC-III database. It divides the patients into three research categories: no readmission, readmission within 30 days, and readmission after 30 days, to predict the readmission of patients. We propose the HR-BGCN model to predict the readmission of patients. First, we use the Adaptive-TMix to improve the prediction indicators of a few categories and reduce the impact of unbalanced categories. Then, the knowledge-informed graph attention mechanism is proposed. By introducing a document-level explicit diagram structure, the coding ability of graph node features is significantly improved. The paragraph-level representation obtained through graph learning is combined with the context token-level representation of BERT, and finally, the multi-classification task is carried out. We also compare several typical graph learning classification models to verify the model's effectiveness, such as the IA-GCN model, GAT model, etc. The results show that the average F1 score of the HR-BGCN model proposed in this paper for 30-day readmission of heart failure patients is 88.26%, and the average accuracy is 90.47%. The HR-BGCN model is significantly better than the graph learning classification model for predicting heart failure readmission. It can help doctors predict the 30-day readmission of patients, then reduce the readmission rate of patients.

Graph global attention network with memory: A deep learning approach for fake news detection

With the proliferation of social media, the detection of fake news has become a critical issue that poses a significant threat to society. The dissemination of fake information can lead to social harm and damage the credibility of information. To address this issue, deep learning has emerged as a promising approach, especially with the development of Natural Language Processing (NLP). This study introduces a novel approach called Graph Global Attention Network with Memory (GANM) for detecting fake news. This approach leverages NLP techniques to encode nodes with news context and user content. It employs three graph convolutional networks to extract informative features from the news propagation network and aggregates endogenous and exogenous user information. This methodology aims to address the challenge of identifying fake news within the context of social media. Innovatively, the GANM combines two strategies. First, a novel global attention mechanism with memory is employed in the GANM to learn the structural homogeneity of news propagation networks, which is the attention mechanism of a single graph with a history of all graphs. Second, we design a module for partial key information learning aggregation to emphasize the acquisition of partial key information in the graph and merge node-level embeddings with graph-level embeddings into fine-grained joint information. Our proposed method provides a new direction in news detection research with a combination of global and partial information and achieves promising performance on real-world datasets.

Prediction of lncRNA and disease associations based on residual graph convolutional networks with attention mechanism

LncRNAs are non-coding RNAs with a length of more than 200 nucleotides. More and more evidence shows that lncRNAs are inextricably linked with diseases. To make up for the shortcomings of traditional methods, researchers began to collect relevant biological data in the database and used bioinformatics prediction tools to predict the associations between lncRNAs and diseases, which greatly improved the efficiency of the study. To improve the prediction accuracy of current methods, we propose a new lncRNA-disease associations prediction method with attention mechanism, called ResGCN-A. Firstly, we integrated lncRNA functional similarity, lncRNA Gaussian interaction profile kernel similarity, disease semantic similarity, and disease Gaussian interaction profile kernel similarity to obtain lncRNA comprehensive similarity and disease comprehensive similarity. Secondly, the residual graph convolutional network was used to extract the local features of lncRNAs and diseases. Thirdly, the new attention mechanism was used to assign the weight of the above features to further obtain the potential features of lncRNAs and diseases. Finally, the training set required by the Extra-Trees classifier was obtained by concatenating potential features, and the potential associations between lncRNAs and diseases were obtained by the trained Extra-Trees classifier. ResGCN-A combines the residual graph convolutional network with the attention mechanism to realize the local and global features fusion of lncRNA and diseases, which is beneficial to obtain more accurate features and improve the prediction accuracy. In the experiment, ResGCN-A was compared with five other methods through 5-fold cross-validation. The results show that the AUC value and AUPR value obtained by ResGCN-A are 0.9916 and 0.9951, which are superior to the other five methods. In addition, case studies and robustness evaluation have shown that ResGCN-A is an effective method for predicting lncRNA-disease associations. The source code for ResGCN-A will be available at https://github.com/Wangxiuxiun/ResGCN-A .

Identifying promoter and enhancer sequences by graph convolutional networks

Identification of promoters, enhancers, and their interactions helps understand genetic regulation. This study proposes a graph-based semi-supervised learning model (GCN4EPI) for the enhancer-promoter classification problem. We adopt a graph convolutional network (GCN) architecture to integrate interaction information with sequence features. Nodes of the constructed graph hold word embeddings of DNA sequences while edges hold the Enhancer-Promoter Interaction (EPI) information. By means of semi-supervised learning, much less data (16%) and time are needed in model training. Comparisons on a benchmark dataset of six human cell lines show that the proposed approach outperforms the state-of-the-art methods by a large margin (10% higher F1 score) and has the fastest training time (up to 3 times). Moreover, GCN4EPI's performance on cross-cell line data is also better than the baselines (3% higher F1 score). Our qualitative analyses with graph explainability models prove that GCN4EPI learns from both text and graph structure. The results suggest that integrating interaction information with sequence features improves predictive performance and compensates for the number of training instances.

Lower and upper bounds for numbers of linear regions of graph convolutional networks

Graph neural networks (GNNs) have become a popular choice for analyzing graph data in the last few years, and characterizing their expressiveness has become an active area of research. One popular measure of expressiveness is the number of linear regions in neural networks with piecewise linear activations. In this paper, we present estimates for the number of linear regions in classic graph convolutional networks (GCNs) with one layer and multiple-layer scenarios and ReLU activation function. We derive an optimal upper bound for the maximum number of linear regions for one-layer GCNs and upper and lower bounds for multi-layer GCNs. Our simulated results suggest that the true maximum number of linear regions is likely closer to our estimated lower bound. These findings indicate that multi-layer GCNs have exponentially greater expressivity than one-layer GCNs per parameter, implying that deeper GCNs are more expressive than their shallow counterparts.

DDI-GCN: Drug-drug interaction prediction via explainable graph convolutional networks

Drug-drug interactions (DDI) may lead to unexpected side effects, which is a growing concern in both academia and industry. Many DDIs have been reported, but the underlying mechanisms are not well understood. Predicting and understanding DDIs can help researchers to improve drug safety and protect patient health. Here, we introduce DDI-GCN, a method that utilizes graph convolutional networks (GCN) to predict DDIs based on chemical structures. We demonstrate that this method achieves state-of-the-art prediction performance on the independent hold-out set. It can also provide visualization of structural features associated with DDIs, which can help us to study the underlying mechanisms. To make it easy and accessible to use, we developed a web server for DDI-GCN, which is freely available at http://wengzq-lab.cn/ddi/.

A framework for predicting odor threshold values of perfumes by scientific machine learning and transfer learning

Knowledge of odor thresholds is very important for the perfume industry. Due to the difficulty associated with measuring odor thresholds, empirical models capable of estimating these values can be an invaluable contribution to the field. This work developed a framework based on scientific machine learning strategies. A transfer learning-based strategy was devised, where information from a graph convolutional network predicting semantic odor descriptors was used as input data for the feedforward neural network responsible for estimating odor thresholds for chemical substances based on their molecular structures. The predictive performance of this model was compared to a benchmark odor threshold prediction model based on molecular structures that did not utilize transfer learning. Furthermore, the prediction was compared to a correlation previously proposed in the literature and a dummy regressor. Results demonstrated that the transfer learning-based strategy displayed a better predictive performance, suggesting this technique can be useful for predicting odor thresholds.

Shall we always use hydraulic models? A graph neural network metamodel for water system calibration and uncertainty assessment

Representing reality in a numerical model is complex. Conventionally, hydraulic models of water distribution networks are a tool for replicating water supply system behaviour through simulation by means of approximation of physical equations. A calibration process is mandatory to achieve plausible simulation results. However, calibration is affected by a set of intrinsic uncertainty sources, mainly related to the lack of system knowledge. This paper proposes a breakthrough approach for calibrating hydraulic models through a graph machine learning approach. The main idea is to create a graph neural network metamodel to estimate the network behaviour based on a limited number of monitoring sensors. Once the flows and pressures of the entire network have been estimated, a calibration is carried out to obtain the set of hydraulic parameters that best approximates the metamodel. Through this process, it is possible to estimate the uncertainty that is transferred from the few available measurements to the final hydraulic model. The paper sparks a discussion to assess under what circumstances a graph-based metamodel might be a solution for water network analysis.

Graph convolutional network with tree-guided anisotropic message passing

Graph Convolutional Networks (GCNs) with naive message passing mechanisms have limited performance due to the isotropic aggregation strategy. To remedy this drawback, some recent works focus on how to design anisotropic aggregation strategies with tricks on feature mapping or structure mining. However, these models still suffer from the low ability of expressiveness and long-range modeling for the needs of high performance in practice. To this end, this paper proposes a tree-guided anisotropic GCN, which applies an anisotropic aggregation strategy with competitive expressiveness and a large receptive field. Specifically, the anisotropic aggregation is decoupled into two stages. The first stage is to establish the path of the message passing on a tree-like hypergraph consisting of substructures. The second one is to aggregate the messages with constrained intensities by employing an effective gating mechanism. In addition, a novel anisotropic readout mechanism is constructed to generate representative and discriminative graph-level features for downstream tasks. Our model outperforms baseline methods and recent works on several synthetic benchmarks and datasets from different real-world tasks. In addition, extensive ablation studies and theoretical analyses indicate the effectiveness of our proposed method.

Graph Spatio-Temporal Networks for Manufacturing Sales Forecast and Prevention Policies in Pandemic Era

Worldwide manufacturing industries are significantly affected by COVID-19 pandemic because of their production characteristics with low-cost country sourcing, globalization, and inventory level. To analyze the correlated time series, spatial-temporal model becomes more attractive, and the graph convolution network (GCN) is also commonly used to provide more information to the nodes and its neighbors in the graph. Recently, attention-adjusted graph spatio-temporal network (AGSTN) was proposed to address the problem of pre-defined graph in GCN by combining multi-graph convolution and attention adjustment to learn spatial and temporal correlations over time. However, AGSTN may show potential problem with limited small non-sensor data; particularly, convergence issue. This study proposes several variants of AGSTN and applies them to non-sensor data. We suggest data augmentation and regularization techniques such as edge selection, time series decomposition, prevention policies to improve AGSTN. An empirical study of worldwide manufacturing industries in pandemic era was conducted to validate the proposed variants. The results show that the proposed variants significantly improve the prediction performance at least around 20% on mean squared error (MSE) and convergence problem.

An extensible hierarchical graph convolutional network for early Alzheimer's disease identification

BACKGROUND AND OBJECTIVE

For early identification of Alzheimer's disease (AD) based on multi-modal magnetic resonance imaging (MRI) data, it is important to make comprehensive use of image features and non-image information to analyze the gray matter atrophy and the structural/functional connectivity abnormalities for different courses of AD.

METHODS

In this study, we propose an extensible hierarchical graph convolutional network (EH-GCN) for early AD identification. Based on the extracted image features from multi-modal MRI data using the presented multi-branch residual network (ResNet), the brain regions-of-interests (ROIs) based GCN is built to extract structural and functional connectivity features between different ROIs of the brain. In order to further improve the performance of AD identification, an optimized spatial GCN is proposed as convolution operator in the population-based GCN to avoid rebuilding the graph network and take advantage of relationships between subjects. Finally, the proposed EH-GCN is built by embedding the image features and internal brain connectivity features into the spatial population-based GCN, which provides an extensible way to improve early AD identification performance by adding imaging features and non-image information from multi-modal data.

RESULTS

Experiments are performed on two datasets, which illustrate the effectiveness of the extracted structural/functional connectivity features and the high computational efficiency of the proposed method. The classification accuracy of AD vs NC, AD vs MCI and MCI vs NC classification tasks reaches 88.71%, 82.71% and 79.68% respectively. The extracted connectivity features between ROIs indicate that functional abnormalities are earlier than gray matter atrophy and abnormalities of structural connections, which is consistent with the clinical manifestations. The proposed method allows for the addition of other modal image features and non-image information from multi-modal data to continuously improve the performance of clinical data analysis.

CONCLUSIONS

The proposed method can help us comprehensively analyze the role of gray matter atrophy, the damage of white matter nerve fiber tracts and the degradation of functional connectivity for different courses of AD, which could be useful for further extraction of clinical biomarkers for early AD identification.

[Research on classification method of multimodal magnetic resonance images of Alzheimer's disease based on generalized convolutional neural networks]

Alzheimer's disease (AD) is a progressive and irreversible neurodegenerative disease. Neuroimaging based on magnetic resonance imaging (MRI) is one of the most intuitive and reliable methods to perform AD screening and diagnosis. Clinical head MRI detection generates multimodal image data, and to solve the problem of multimodal MRI processing and information fusion, this paper proposes a structural and functional MRI feature extraction and fusion method based on generalized convolutional neural networks (gCNN). The method includes a three-dimensional residual U-shaped network based on hybrid attention mechanism (3D HA-ResUNet) for feature representation and classification for structural MRI, and a U-shaped graph convolutional neural network (U-GCN) for node feature representation and classification of brain functional networks for functional MRI. Based on the fusion of the two types of image features, the optimal feature subset is selected based on discrete binary particle swarm optimization, and the prediction results are output by a machine learning classifier. The validation results of multimodal dataset from the AD Neuroimaging Initiative (ADNI) open-source database show that the proposed models have superior performance in their respective data domains. The gCNN framework combines the advantages of these two models and further improves the performance of the methods using single-modal MRI, improving the classification accuracy and sensitivity by 5.56% and 11.11%, respectively. In conclusion, the gCNN-based multimodal MRI classification method proposed in this paper can provide a technical basis for the auxiliary diagnosis of Alzheimer's disease.

Multi-relational graph convolutional networks: Generalization guarantees and experiments

The class of multi-relational graph convolutional networks (MRGCNs) is a recent extension of standard graph convolutional networks (GCNs) to handle heterogenous graphs with multiple types of relationships. MRGCNs have been shown to yield results superior than traditional GCNs in various machine learning tasks. The key idea is to introduce a new kind of convolution operated on tensors that can effectively exploit correlations exhibited in multiple relationships. The main objective of this paper is to analyze the algorithmic stability and generalization guarantees of MRGCNs to confirm the usefulness of MRGCNs. Our contributions are of three folds. First, we develop a matrix representation of various tensor operations underneath MRGCNs to simplify the analysis significantly. Next, we prove the uniform stability of MRGCNs and deduce the convergence of the generalization gap to support the usefulness of MRGCNs. The analysis sheds lights on the design of MRGCNs, for instance, how the data should be scaled to achieve the uniform stability of the learning process. Finally, we provide experimental results to demonstrate the stability results.

Exploring interpretable graph convolutional networks for autism spectrum disorder diagnosis

PURPOSE

Finding the biomarkers associated with autism spectrum disorder (ASD) is helpful for understanding the underlying roots of the disorder and can lead to earlier diagnosis and more targeted treatments. In essence, we are faced with two challenges (i) how to learn a node representation and a clean graph structure from original graph data with high dimensionality and (ii) how to jointly model the procedure of node representation learning, structure learning and graph classification.

METHODS

We propose FSL-BrainNet, an interpretable graph convolution network (GCN) model for jointly Learning of node Features and clean Structures in brain networks for automatic brain network classification and interpretation. We formulate an end-to-end trainable and interpretable framework for graph classification and biomarkers (salient brain regions and potential subnetworks) identification.

RESULTS

The experimental results on the ABIDE dataset show that our proposed methods not only achieve improved prediction performance compared with the state-of-the-art methods, but also find a compact set of highly suggestive biomarkers including relevant brain regions and subnetworks to ASD.

CONCLUSION

Through node feature learning and structure learning, our model can simultaneously select important brain regions and identify subnetworks.

RNN-combined graph convolutional network with multi-feature fusion for tuberculosis cavity segmentation

Tuberculosis is a common infectious disease in the world. Tuberculosis cavities are common and an important imaging signs in tuberculosis. Accurate segmentation of tuberculosis cavities has practical significance for indicating the activity of lesions and guiding clinical treatment. However, this task faces challenges such as blurred boundaries, irregular shapes, different location and size of lesions and similar structures on computed tomography (CT) to other lung diseases or tissues. To overcome these problems, we propose a novel RNN-combined graph convolutional network (R2GCN) method, which integrates the bidirectional recurrent network (BRN) and graph convolution network (GCN) modules. First, feature extraction is performed on the input image by VGG-16 or ResNet-50 to obtain the feature map. The feature map is then used as the input of the two modules. On the one hand, we adopt the BRN to retrieve contextual information from the feature map. On the other hand, we take the vector for each location in the feature map as input nodes and utilize GCN to extract node topology information. Finally, two types of features obtained fuse together. Our strategy can not only make full use of node correlations and differences, but also obtain more precise segmentation boundaries. Extensive experiments on CT images of cavitary patients with tuberculosis show that our proposed method achieves the best segmentation accuracy than compared segmentation methods. Our method can be used for the diagnosis of tuberculosis cavity and the evaluation of tuberculosis cavity treatment.

Multi-graph Fusion Graph Convolutional Networks with pseudo-label supervision

Graph convolutional networks (GCNs) have become a popular tool for learning unstructured graph data due to their powerful learning ability. Many researchers have been interested in fusing topological structures and node features to extract the correlation information for classification tasks. However, it is inadequate to integrate the embedding from topology and feature spaces to gain the most correlated information. At the same time, most GCN-based methods assume that the topology graph or feature graph is compatible with the properties of GCNs, but this is usually not satisfied since meaningless, missing, or even unreal edges are very common in actual graphs. To obtain a more robust and accurate graph structure, we intend to construct an adaptive graph with topology and feature graphs. We propose Multi-graph Fusion Graph Convolutional Networks with pseudo-label supervision (MFGCN), which learn a connected embedding by fusing the multi-graphs and node features. We can obtain the final node embedding for semi-supervised node classification by propagating node features over multi-graphs. Furthermore, to alleviate the problem of labels missing in semi-supervised classification, a pseudo-label generation mechanism is proposed to generate more reliable pseudo-labels based on the similarity of node features. Extensive experiments on six benchmark datasets demonstrate the superiority of MFGCN over state-of-the-art classification methods.

KAGN:knowledge-powered attention and graph convolutional networks for social media rumor detection

Rumor posts have received substantial attention with the rapid development of online and social media platforms. The automatic detection of rumor from posts has emerged as a major concern for the general public, the government, and social media platforms. Most existing methods focus on the linguistic and semantic aspects of posts content, while ignoring knowledge entities and concepts hidden within the article which facilitate rumor detection. To address these limitations, in this paper, we propose a novel end-to-end attention and graph-based neural network model (KAGN), which incorporates external knowledge from the knowledge graphs to detect rumor. Specifically, given the post's sparse and ambiguous semantics, we identify entity mentions in the post's content and link them to entities and concepts in the knowledge graphs, which serve as complementary semantic information for the post text. To effectively inject external knowledge into textual representations, we develop a knowledge-aware attention mechanism to fuse local knowledge. Additionally, we construct a graph consisting of posts texts, entities, and concepts, which is fed to graph convolutional networks to explore long-range knowledge through graph structure. Our proposed model can therefore detect rumor by combining semantic-level and knowledge-level representations of posts. Extensive experiments on four publicly available real-world datasets show that KAGN outperforms or is comparable to other state-of-the-art methods, and also validate the effectiveness of knowledge.

Graph-Based Disease Prediction in Neuroimaging: Investigating the Impact of Feature Selection

In biomedical machine learning, data often appear in the form of graphs. Biological systems such as protein interactions and ecological or brain networks are instances of applications that benefit from graph representations. Geometric deep learning is an arising field of techniques that has extended deep neural networks to non-Euclidean domains such as graphs. In particular, graph convolutional neural networks have achieved advanced performance in semi-supervised learning in those domains. Over the last years, these methods have gained traction in neuroscience as they could be the key to a deeper understanding in clinical diagnosis at the systems or network level (for an individual brain but also for across a cohort of subjects). As a proof-of-principle, we study and validate a previous implementation of graph-based semi-supervised classification using a ridge classifier and graph convolutional neural networks. The models are trained on population graphs that integrate imaging and phenotypic information. Our analysis employs neuroimaging data of structural and functional connectivity for prediction of neurodevelopmental and neurodegenerative disorders. Here, we particularly study the effect of different strategies to reduce the dimensionality of the neuroimaging features on the graph nodes on the classification performance.

A novel candidate disease gene prioritization method using deep graph convolutional networks and semi-supervised learning

Background

Selecting and prioritizing candidate disease genes is necessary before conducting laboratory studies as identifying disease genes from a large number of candidate genes using laboratory methods, is a very costly and time-consuming task. There are many machine learning-based gene prioritization methods. These methods differ in various aspects including the feature vectors of genes, the used datasets with different structures, and the learning model. Creating a suitable feature vector for genes and an appropriate learning model on a variety of data with different and non-Euclidean structures, including graphs, as well as the lack of negative data are very important challenges of these methods. The use of graph neural networks has recently emerged in machine learning and other related fields, and they have demonstrated superior performance for a broad range of problems.

Methods

In this study, a new semi-supervised learning method based on graph convolutional networks is presented using the novel constructing feature vector for each gene. In the proposed method, first, we construct three feature vectors for each gene using terms from the Gene Ontology (GO) database. Then, we train a graph convolution network on these vectors using protein–protein interaction (PPI) network data to identify disease candidate genes. Our model discovers hidden layer representations encoding in both local graph structure as well as features of nodes. This method is characterized by the simultaneous consideration of topological information of the biological network (e.g., PPI) and other sources of evidence. Finally, a validation has been done to demonstrate the efficiency of our method.

Results

Several experiments are performed on 16 diseases to evaluate the proposed method's performance. The experiments demonstrate that our proposed method achieves the best results, in terms of precision, the area under the ROC curve (AUCs), and F1-score values, when compared with eight state-of-the-art network and machine learning-based disease gene prioritization methods.

Conclusion

This study shows that the proposed semi-supervised learning method appropriately classifies and ranks candidate disease genes using a graph convolutional network and an innovative method to create three feature vectors for genes based on the molecular function, cellular component, and biological process terms from GO data.

MHDMF: Prediction of miRNA-disease associations based on Deep Matrix Factorization with Multi-source Graph Convolutional Network

A growing number of works have proved that microRNAs (miRNAs) are a crucial biomarker in diverse bioprocesses affecting various diseases. As a good complement to high-cost wet experiment-based methods, numerous computational prediction methods have sprung up. However, there are still challenges that exist in making effective use of high false-negative associations and multi-source information for finding the potential associations. In this work, we develop an end-to-end computational framework, called MHDMF, which integrates the multi-source information on a heterogeneous network to discover latent disease-miRNA associations. Since high false-negative exist in the miRNA-disease associations, MHDMF utilizes the multi-source Graph Convolutional Network (GCN) to correct the false-negative association by reformulating the miRNA-disease association score matrix. The score matrix reformulation is based on different similarity profiles and known associations between miRNAs, genes, and diseases. Then, MHDMF employs Deep Matrix Factorization (DMF) to predict the miRNA-disease associations based on reformulated miRNA-disease association score matrix. The experimental results show that the proposed framework outperforms highly related comparison methods by a large margin on tasks of miRNA-disease association prediction. Furthermore, case studies suggest that MHDMF could be a convenient and efficient tool and may supply a new way to think about miRNA-disease association prediction.

MVGCNMDA: Multi-view Graph Augmentation Convolutional Network for Uncovering Disease-Related Microbes

MOTIVATION

Exploring the interrelationships between microbes and disease can help microbiologists make decisions and plan treatments. Predicting new microbe-disease associations currently relies on biological experiments and domain knowledge, which is time-consuming and inefficient. Automated algorithms are used to uncover the intrinsic link between microbes and disease. However, due to data noise and inadequate understanding of relevant biology, the efficient prediction of microbe-disease associations is still crucial. This study develops a multi-view graph augmentation convolutional network (MVGCNMDA) to predict potential disease-associated microbes.

METHODS

First, we use two data augmentation methods, edge perturbation and node dropping, to remove the data noise in the preprocessing stage. Second, we calculate Gaussian interaction profile kernel similarity and cosine similarity. Therefore, the Graph Convolutional Network(GCN) can fully use multi-view features. Then, the multi-view features are fed into the multi-attention block to learn the weights of different features adaptively. Finally, the embedding results are obtained using a Convolutional Neural Network (CNN) combiner, and the matrix completion is used to predict the relationship between potential microbes and diseases.

RESULTS

We test our model on the Human microbe-disease Association Database (HMDAD), Disbiome, and the Combined Dataset (Peryton and MicroPhenoDB). The area under PR curve (AUPR), area under ROC curve (AUC), F1 score, and RECALL value are calculated to evaluate the performance of the developed MVGCNMDA. The AUPR is 0.9440, AUC is 0.9428, F1 score is 0.9383, and RECALL value is 0.8858. The experiments show that our model can accurately predict potential microbe-disease associations compared with the state-of-the-art works on the global Leave-One-Out-Cross-Validation (LOOCV) and the fivefold Cross-Validation (fivefold CV). To further verify the effectiveness of the proposed graph data augmentation, we designed five different settings in the ablation study. Furthermore, we present two case studies that validate the prediction of the potential association between microbes and diseases by MVGCNMDA.

Graph convolutional network based virus-human protein-protein interaction prediction for novel viruses

Computational identification of human-virus protein-protein interactions (PHIs) is a worthwhile step towards understanding infection mechanisms. Analysis of the PHI networks is important for the determination of pathogenic diseases. Prediction of these interactions is a popular problem since experimental detection of PHIs is both time-consuming and expensive. The available methods use biological features like amino acid sequences, molecular structure, or biological activities for prediction. Recent studies show that the topological properties of proteins in protein-protein interaction (PPI) networks increase the performance of the predictions. The basic network projections, random-walk-based models, or graph neural networks are used for generating topologically enriched (hybrid) protein embeddings. In this study, we propose a three-stage machine learning pipeline that generates and uses hybrid embeddings for PHI prediction. In the first stage, numerical features are extracted from the amino acid sequences using the Doc2Vec and Byte Pair Encoding method. The amino acid embeddings are used as node features while training a modified GraphSAGE model, which is an improved version of the graph convolutional network. Lastly, the hybrid protein embeddings are used for training a binary interaction classifier model that predicts whether there is an interaction between the given two proteins or not. The proposed method is evaluated with comprehensive experiments to test its functionality and compare it with the state-of-art methods. The experimental results on the benchmark dataset prove the efficiency of the proposed model by having a 3-23% better area under curve (AUC) score than its competitors.

Collaborative learning of graph generation, clustering and classification for brain networks diagnosis

PURPOSE

Accurate diagnosis of autism spectrum disorder (ASD) plays a key role in improving the condition and quality of life for patients. In this study, we mainly focus on ASD diagnosis with functional brain networks (FBNs). The major challenge for brain networks modeling is the high dimensional connectivity in brain networks and limited number of subjects, which hinders the classification capability of graph convolutional networks (GCNs).

METHOD

To alleviate the influence of the limited data and high dimensional connectivity, we introduce a unified three-stage graph learning framework for brain network classification, involving multi-graph clustering, graph generation and graph classification. The framework combining Graph Generation, Clustering and Classification Networks (GraphCGC-Net) enhances the critical connections by multi-graph clustering (MGC) with a supervision scheme, and generates realistic brain networks by simultaneously preserving the global consistent distribution and local topology properties.

RESULTS

To demonstrate the effectiveness of our approach, we evaluate the performance of the proposed method on the Autism Brain Imaging Data Exchange (ABIDE) dataset and conduct extensive experiments on the ASD classification problem. Our proposed method achieves an average accuracy of 70.45% and an AUC of 72.76% on ABIDE. Compared with the traditional GCN model, the proposed GraphCGC-Net obtains 9.3%, and 10.64% improvement in terms of accuracy and AUC metrics, respectively.

CONCLUSION

The comprehensive experiments demonstrate that our GraphCGC-Net is effective for graph classification in brain disorders diagnosis. Moreover, we find that MGC can generate biologically meaningful subnetworks, which is highly consistent with the previous neuroimaging-derived biomarker evidence of ASD. More importantly, the promising results suggest that applying generative adversarial networks (GANs) in brain networks to improve the classification performance is worth further investigation.

HPE-GCN: predicting efficacy of tonic formulae via graph convolutional networks integrating traditionally defined herbal properties

Chinese herbal formulae are the heritage of traditional Chinese medicine (TCM) in treating diseases through thousands of years. The formula function is not just a simple herbal efficacy addition, but produces complex and nonlinear relationships between different herbs and their overall efficacy, which brings challenges to the formula efficacy analysis. In our study, we proposed a model called HPE-GCN that combines graph convolutional networks (GCN) with TCM-defined herbal properties (TCM-HPs) to predict formulae efficacy. In addition, to process the unstructured natural language in the formula text, we proposed a weighting calculation method related to herb frequency and the number of herbs in a formula called Formula-Herb dependence degree (FHDD), to assess the dependency degree of a formula with its herbs. In our research, 214 classic tonic formulae from ancient TCM books such as Synopsis of the Golden Chamber, Jingyue's Complete Works and the Golden Mirror of Medicin were collected as datasets. The performance of HPE-GCN on multi-classification of tonic formulae reached the best result compared with classic machine learning models, such as support vector machine, naive Bayes, logistic regression, gradient boosting decision tree, and K-nearest neighbors. The evaluated index Macro-Precision, Macro-Recall, Macro-F1 of HPE-GCN on the test set were 87.70%, 84.08% and 83.51% respectively, increased by 7.27%, 7.41% and 7.30% respectively from second best compared models. GCN has the advantage of low-dimensional feature expression for herbs and formulae, and is an effective analysis tool for TCM research. HPE-GCN integrates TCM-HPs and fits the complex nonlinear mapping relationship between TCM-HPs and formulae efficacy, which provides new ideas for related research.

MedGCN: Medication recommendation and lab test imputation via graph convolutional networks

Laboratory testing and medication prescription are two of the most important routines in daily clinical practice. Developing an artificial intelligence system that can automatically make lab test imputations and medication recommendations can save costs on potentially redundant lab tests and inform physicians of a more effective prescription. We present an intelligent medical system (named MedGCN) that can automatically recommend the patients' medications based on their incomplete lab tests, and can even accurately estimate the lab values that have not been taken. In our system, we integrate the complex relations between multiple types of medical entities with their inherent features in a heterogeneous graph. Then we model the graph to learn a distributed representation for each entity in the graph based on graph convolutional networks (GCN). By the propagation of graph convolutional networks, the entity representations can incorporate multiple types of medical information that can benefit multiple medical tasks. Moreover, we introduce a cross regularization strategy to reduce overfitting for multi-task training by the interaction between the multiple tasks. In this study, we construct a graph to associate 4 types of medical entities, i.e., patients, encounters, lab tests, and medications, and applied a graph neural network to learn node embeddings for medication recommendation and lab test imputation. we validate our MedGCN model on two real-world datasets: NMEDW and MIMIC-III. The experimental results on both datasets demonstrate that our model can outperform the state-of-the-art in both tasks. We believe that our innovative system can provide a promising and reliable way to assist physicians to make medication prescriptions and to save costs on potentially redundant lab tests.

A dynamic graph convolutional neural network framework reveals new insights into connectome dysfunctions in ADHD

The pathological mechanism of attention deficit hyperactivity disorder (ADHD) is incompletely specified, which leads to difficulty in precise diagnosis. Functional magnetic resonance imaging (fMRI) has emerged as a common neuroimaging technique for studying the brain functional connectome. Most existing methods that have either ignored or simply utilized graph structure, do not fully leverage the potentially important topological information which may be useful in characterizing brain disorders. There is a crucial need for designing novel and efficient approaches which can capture such information. To this end, we propose a new dynamic graph convolutional network (dGCN), which is trained with sparse brain regional connections from dynamically calculated graph features. We also develop a novel convolutional readout layer to improve graph representation. Our extensive experimental analysis demonstrates significantly improved performance of dGCN for ADHD diagnosis compared with existing machine learning and deep learning methods. Visualizations of the salient regions of interest (ROIs) and connectivity based on informative features learned by our model show that the identified functional abnormalities mainly involve brain regions in temporal pole, gyrus rectus, and cerebellar gyri from temporal lobe, frontal lobe, and cerebellum, respectively. A positive correlation was further observed between the identified connectomic abnormalities and ADHD symptom severity. The proposed dGCN model shows great promise in providing a functional network-based precision diagnosis of ADHD and is also broadly applicable to brain connectome-based study of mental disorders.

TE-HI-GCN: An Ensemble of Transfer Hierarchical Graph Convolutional Networks for Disorder Diagnosis

Accurate diagnosis of psychiatric disorders plays a critical role in improving the quality of life for patients and potentially supports the development of new treatments. Graph convolutional networks (GCNs) are shown to be successful in modeling applications with graph structures. However, training an accurate GCNs model for brain networks faces several challenges, including high dimensional and noisy correlation in the brain networks, limited labeled training data, and depth limitation of GCN learning. Generalization and interpretability are important in developing predictive models for clinical diagnosis. To address these challenges, we proposed an ensemble framework involving hierarchical GCN and transfer learning for sparse brain networks, which allows GCN to capture the intrinsic correlation among the subjects and domains, to improve the network embedding learning for disease diagnosis. Extensive experiments on two real medical clinical applications: diagnosis of Autism spectrum disorder (ASD) and diagnosis of Alzheimer's disease (AD) on both the ADNI and ABIDE databases, showing the effectiveness of the proposed framework. We achieved state-of-the-art accuracy and AUC for AD/MCI and ASD/NC (Normal control) classification in comparison with studies that used functional connectivity as features or GCN models. The proposed TE-HI-GCN model achieves the best classification performance, leading to about 27.93% (31.38%) improvement for ASD and 16.86% (44.50%) for AD in terms of accuracy and AUC compared with the traditional GCN model. Moreover, the obtained clustering results show high correspondence with the previous neuroimaging derived evidence of within and between-networks biomarkers for ASD. The discovered subnetworks are used as evidence for the proposed TE-HI-GCN model. Furthermore, this work is the first attempt of transfer learning on the two related disorder domains to uncover the correlation among the two diseases with a transfer learning scheme.

Multi-view graph embedding clustering network: Joint self-supervision and block diagonal representation

Multi-view clustering has become an active topic in artificial intelligence. Yet, similar investigation for graph-structured data clustering has been absent so far. To fill this gap, we present a Multi-View Graph embedding Clustering network (MVGC). Specifically, unlike traditional multi-view construction methods, which are only suitable to describe Euclidean structure data, we leverage Euler transform to augment the node attribute, as a new view descriptor, for non-Euclidean structure data. Meanwhile, we impose block diagonal representation constraint, which is measured by the ℓ_1,2-norm, on self-expression coefficient matrix to well explore the cluster structure. By doing so, the learned view-consensus coefficient matrix well encodes the discriminative information. Moreover, we make use of the learned clustering labels to guide the learnings of node representation and coefficient matrix, where the latter is used in turn to conduct the subsequent clustering. In this way, clustering and representation learning are seamlessly connected, with the aim to achieve better clustering performance. Extensive experimental results indicate that MVGC is superior to 11 state-of-the-art methods on four benchmark datasets. In particular, MVGC achieves an Accuracy of 96.17% (53.31%) on the ACM (IMDB) dataset, which is an up to 2.85% (1.97%) clustering performance improvement compared with the strongest baseline.

RA-GCN: Graph convolutional network for disease prediction problems with imbalanced data

Disease prediction is a well-known classification problem in medical applications. Graph Convolutional Networks (GCNs) provide a powerful tool for analyzing the patients' features relative to each other. This can be achieved by modeling the problem as a graph node classification task, where each node is a patient. Due to the nature of such medical datasets, class imbalance is a prevalent issue in the field of disease prediction, where the distribution of classes is skewed. When the class imbalance is present in the data, the existing graph-based classifiers tend to be biased towards the major class(es) and neglect the samples in the minor class(es). On the other hand, the correct diagnosis of the rare positive cases (true-positives) among all the patients is vital in a healthcare system. In conventional methods, such imbalance is tackled by assigning appropriate weights to classes in the loss function which is still dependent on the relative values of weights, sensitive to outliers, and in some cases biased towards the minor class(es). In this paper, we propose a Re-weighted Adversarial Graph Convolutional Network (RA-GCN) to prevent the graph-based classifier from emphasizing the samples of any particular class. This is accomplished by associating a graph-based neural network to each class, which is responsible for weighting the class samples and changing the importance of each sample for the classifier. Therefore, the classifier adjusts itself and determines the boundary between classes with more attention to the important samples. The parameters of the classifier and weighting networks are trained by an adversarial approach. We show experiments on synthetic and three publicly available medical datasets. Our results demonstrate the superiority of RA-GCN compared to recent methods in identifying the patient's status on all three datasets. The detailed analysis of our method is provided as quantitative and qualitative experiments on synthetic datasets.

A deep-learning approach for direct whole-heart mesh reconstruction

Automated construction of surface geometries of cardiac structures from volumetric medical images is important for a number of clinical applications. While deep-learning-based approaches have demonstrated promising reconstruction precision, these approaches have mostly focused on voxel-wise segmentation followed by surface reconstruction and post-processing techniques. However, such approaches suffer from a number of limitations including disconnected regions or incorrect surface topology due to erroneous segmentation and stair-case artifacts due to limited segmentation resolution. We propose a novel deep-learning-based approach that directly predicts whole heart surface meshes from volumetric CT and MR image data. Our approach leverages a graph convolutional neural network to predict deformation on mesh vertices from a pre-defined mesh template to reconstruct multiple anatomical structures in a 3D image volume. Our method demonstrated promising performance of generating whole heart reconstructions with as good or better accuracy than prior deep-learning-based methods on both CT and MR data. Furthermore, by deforming a template mesh, our method can generate whole heart geometries with better anatomical consistency and produce high-resolution geometries from lower resolution input image data. Our method was also able to produce temporally-consistent surface mesh predictions for heart motion from CT or MR cine sequences, and therefore can potentially be applied for efficiently constructing 4D whole heart dynamics. Our code and pre-trained networks are available at https://github.com/fkong7/MeshDeformNet.

ABCDE: Approximating Betweenness-Centrality ranking with progressive-DropEdge

Betweenness-centrality is a popular measure in network analysis that aims to describe the importance of nodes in a graph. It accounts for the fraction of shortest paths passing through that node and is a key measure in many applications including community detection and network dismantling. The computation of betweenness-centrality for each node in a graph requires an excessive amount of computing power, especially for large graphs. On the other hand, in many applications, the main interest lies in finding the top-k most important nodes in the graph. Therefore, several approximation algorithms were proposed to solve the problem faster. Some recent approaches propose to use shallow graph convolutional networks to approximate the top-k nodes with the highest betweenness-centrality scores. This work presents a deep graph convolutional neural network that outputs a rank score for each node in a given graph. With careful optimization and regularization tricks, including an extended version of DropEdge which is named Progressive-DropEdge, the system achieves better results than the current approaches. Experiments on both real-world and synthetic datasets show that the presented algorithm is an order of magnitude faster in inference and requires several times fewer resources and time to train.

Dual-Sampling Attention Pooling for Graph Neural Networks on 3D Mesh

Mesh is an essential and effective data representation of a 3D shape. The 3D mesh segmentation is a fundamental task in computer vision and graphics. It has recently been realized through a multi-scale deep learning framework, whose sampling methods are of key significance. Rarely do the previous sampling methods consider the receptive field contour of vertex, leading to loss in scale consistency of the vertex feature. Meanwhile, uniform sampling can ensure the utmost uniformity of the vertex distribution of the sampled mesh. Consequently, to efficiently improve the scale consistency of vertex features, uniform sampling was first used in this study to construct a multi-scale mesh hierarchy. In order to address the issue on uniform sampling, namely, the smoothing effect, vertex clustering sampling was used because it can preserve the geometric structure, especially the edge information. With the merits of these two sampling methods combined, more and complete information on the 3D shape can be acquired. Moreover, we adopted the attention mechanism to better realize the cross-scale shape feature transfer. According to the attention mechanism, shape feature transfer between different scales can be realized by the construction of a novel graph structure. On this basis, we propose dual-sampling attention pooling for graph neural networks on 3D mesh. According to experiments on three datasets, the proposed methods are highly competitive.

Deep attributed graph clustering with self-separation regularization and parameter-free cluster estimation

Detecting clusters over attributed graphs is a fundamental task in the graph analysis field. The goal is to partition nodes into dense clusters based on both their attributes and structures. Modern graph neural networks provide facilitation to jointly capture the above information in attributed graphs with a feature aggregation manner, and have achieved great success in attributed graph clustering. However, existing methods mainly focus on capturing the proximity information in graphs and often fail to learn cluster-friendly features during the training of models. Besides, similar to many deep clustering frameworks, current methods based on graph neural networks require a preassigned cluster number before estimating the clusters. To address these limitations, we propose in this paper a deep attributed clustering method based on self-separated graph neural networks and parameter-free cluster estimation. First, to learn cluster-friendly features, we jointly optimize a jumping graph convolutional auto-encoder with a self-separation regularizer, which learns clusters with changing sizes while keeping dense intra-cluster structures and sparse inter structures. Second, an additional softmax auto-encoder is trained to determine the natural cluster number from the data. The hidden units capture cluster structures and can be used to estimate the number of clusters. Extensive experiments show the effectiveness of the proposed model.

Joint fully convolutional and graph convolutional networks for weakly-supervised segmentation of pathology images

Tissue/region segmentation of pathology images is essential for quantitative analysis in digital pathology. Previous studies usually require full supervision (e.g., pixel-level annotation) which is challenging to acquire. In this paper, we propose a weakly-supervised model using joint Fully convolutional and Graph convolutional Networks (FGNet) for automated segmentation of pathology images. Instead of using pixel-wise annotations as supervision, we employ an image-level label (i.e., foreground proportion) as weakly-supervised information for training a unified convolutional model. Our FGNet consists of a feature extraction module (with a fully convolutional network) and a classification module (with a graph convolutional network). These two modules are connected via a dynamic superpixel operation, making the joint training possible. To achieve robust segmentation performance, we propose to use mutable numbers of superpixels for both training and inference. Besides, to achieve strict supervision, we employ an uncertainty range constraint in FGNet to reduce the negative effect of inaccurate image-level annotations. Compared with fully-supervised methods, the proposed FGNet achieves competitive segmentation results on three pathology image datasets (i.e., HER2, KI67, and H&E) for cancer region segmentation, suggesting the effectiveness of our method. The code is made publicly available at https://github.com/zhangjun001/FGNet.

FastGAE: Scalable graph autoencoders with stochastic subgraph decoding

Graph autoencoders (AE) and variational autoencoders (VAE) are powerful node embedding methods, but suffer from scalability issues. In this paper, we introduce FastGAE, a general framework to scale graph AE and VAE to large graphs with millions of nodes and edges. Our strategy, based on an effective stochastic subgraph decoding scheme, significantly speeds up the training of graph AE and VAE while preserving or even improving performances. We demonstrate the effectiveness of FastGAE on various real-world graphs, outperforming the few existing approaches to scale graph AE and VAE by a wide margin.

MutagenPred-GCNNs: A Graph Convolutional Neural Network-Based Classification Model for Mutagenicity Prediction with Data-Driven Molecular Fingerprints

An important task in the early stage of drug discovery is the identification of mutagenic compounds. Mutagenicity prediction models that can interpret relationships between toxicological endpoints and compound structures are especially favorable. In this research, we used an advanced graph convolutional neural network (GCNN) architecture to identify the molecular representation and develop predictive models based on these representations. The predictive model based on features extracted by GCNNs can not only predict the mutagenicity of compounds but also identify the structure alerts in compounds. In fivefold cross-validation and external validation, the highest area under the curve was 0.8782 and 0.8382, respectively; the highest accuracy (Q) was 80.98% and 76.63%, respectively; the highest sensitivity was 83.27% and 78.92%, respectively; and the highest specificity was 78.83% and 76.32%, respectively. Additionally, our model also identified some toxicophores, such as aromatic nitro, three-membered heterocycles, quinones, and nitrogen and sulfur mustard. These results indicate that GCNNs could learn the features of mutagens effectively. In summary, we developed a mutagenicity classification model with high predictive performance and interpretability based on a data-driven molecular representation trained through GCNNs.

Combining sequence and network information to enhance protein-protein interaction prediction

Background

Protein–protein interactions (PPIs) are of great importance in cellular systems of organisms, since they are the basis of cellular structure and function and many essential cellular processes are related to that. Most proteins perform their functions by interacting with other proteins, so predicting PPIs accurately is crucial for understanding cell physiology.

Results

Recently, graph convolutional networks (GCNs) have been proposed to capture the graph structure information and generate representations for nodes in the graph. In our paper, we use GCNs to learn the position information of proteins in the PPIs networks graph, which can reflect the properties of proteins to some extent. Combining amino acid sequence information and position information makes a stronger representation for protein, which improves the accuracy of PPIs prediction.

Conclusion

In previous research methods, most of them only used protein amino acid sequence as input information to make predictions, without considering the structural information of PPIs networks graph. We first time combine amino acid sequence information and position information to make representations for proteins. The experimental results indicate that our method has strong competitiveness compared with several sequence-based methods.

GCNG: graph convolutional networks for inferring gene interaction from spatial transcriptomics data

Most methods for inferring gene-gene interactions from expression data focus on intracellular interactions. The availability of high-throughput spatial expression data opens the door to methods that can infer such interactions both within and between cells. To achieve this, we developed Graph Convolutional Neural networks for Genes (GCNG). GCNG encodes the spatial information as a graph and combines it with expression data using supervised training. GCNG improves upon prior methods used to analyze spatial transcriptomics data and can propose novel pairs of extracellular interacting genes. The output of GCNG can also be used for downstream analysis including functional gene assignment.Supporting website with software and data: https://github.com/xiaoyeye/GCNG .

Identification of early mild cognitive impairment using multi-modal data and graph convolutional networks

BACKGROUND

The identification of early mild cognitive impairment (EMCI), which is an early stage of Alzheimer's disease (AD) and is associated with brain structural and functional changes, is still a challenging task. Recent studies show great promises for improving the performance of EMCI identification by combining multiple structural and functional features, such as grey matter volume and shortest path length. However, extracting which features and how to combine multiple features to improve the performance of EMCI identification have always been a challenging problem. To address this problem, in this study we propose a new EMCI identification framework using multi-modal data and graph convolutional networks (GCNs). Firstly, we extract grey matter volume and shortest path length of each brain region based on automated anatomical labeling (AAL) atlas as feature representation from T1w MRI and rs-fMRI data of each subject, respectively. Then, in order to obtain features that are more helpful in identifying EMCI, a common multi-task feature selection method is applied. Afterwards, we construct a non-fully labelled subject graph using imaging and non-imaging phenotypic measures of each subject. Finally, a GCN model is adopted to perform the EMCI identification task.

RESULTS

Our proposed EMCI identification method is evaluated on 210 subjects, including 105 subjects with EMCI and 105 normal controls (NCs), with both T1w MRI and rs-fMRI data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. Experimental results show that our proposed framework achieves an accuracy of 84.1% and an area under the receiver operating characteristic (ROC) curve (AUC) of 0.856 for EMCI/NC classification. In addition, by comparison, the accuracy and AUC values of our proposed framework are better than those of some existing methods in EMCI identification.

CONCLUSION

Our proposed EMCI identification framework is effective and promising for automatic diagnosis of EMCI in clinical practice.

Interpretation of Brain Morphology in Association to Alzheimer's Disease Dementia Classification Using Graph Convolutional Networks on Triangulated Meshes

We propose a mesh-based technique to aid in the classification of Alzheimer's disease dementia (ADD) using mesh representations of the cortex and subcortical structures. Deep learning methods for classification tasks that utilize structural neuroimaging often require extensive learning parameters to optimize. Frequently, these approaches for automated medical diagnosis also lack visual interpretability for areas in the brain involved in making a diagnosis. This work: (a) analyzes brain shape using surface information of the cortex and subcortical structures, (b) proposes a residual learning framework for state-of-the-art graph convolutional networks which offer a significant reduction in learnable parameters, and (c) offers visual interpretability of the network via class-specific gradient information that localizes important regions of interest in our inputs. With our proposed method leveraging the use of cortical and subcortical surface information, we outperform other machine learning methods with a 96.35% testing accuracy for the ADD vs. healthy control problem. We confirm the validity of our model by observing its performance in a 25-trial Monte Carlo cross-validation. The generated visualization maps in our study show correspondences with current knowledge regarding the structural localization of pathological changes in the brain associated to dementia of the Alzheimer's type.

Graph convolutional networks: a comprehensive review

Graphs naturally appear in numerous application domains, ranging from social analysis, bioinformatics to computer vision. The unique capability of graphs enables capturing the structural relations among data, and thus allows to harvest more insights compared to analyzing data in isolation. However, it is often very challenging to solve the learning problems on graphs, because (1) many types of data are not originally structured as graphs, such as images and text data, and (2) for graph-structured data, the underlying connectivity patterns are often complex and diverse. On the other hand, the representation learning has achieved great successes in many areas. Thereby, a potential solution is to learn the representation of graphs in a low-dimensional Euclidean space, such that the graph properties can be preserved. Although tremendous efforts have been made to address the graph representation learning problem, many of them still suffer from their shallow learning mechanisms. Deep learning models on graphs (e.g., graph neural networks) have recently emerged in machine learning and other related areas, and demonstrated the superior performance in various problems. In this survey, despite numerous types of graph neural networks, we conduct a comprehensive review specifically on the emerging field of graph convolutional networks, which is one of the most prominent graph deep learning models. First, we group the existing graph convolutional network models into two categories based on the types of convolutions and highlight some graph convolutional network models in details. Then, we categorize different graph convolutional networks according to the areas of their applications. Finally, we present several open challenges in this area and discuss potential directions for future research.