Zhengwang Xia
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Zhengwang Xia has a total of 31 co-authors. Here are the top co-authors by number of shared publications:
- Jianfeng Lu
Nanjing University of Science and Technology
8 shared publications - Saqib Mamoon
School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China
4 shared publications - Amani Alfakih
School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China
3 shared publications - Zhuqing Jiao
School of Computer Science and Artificial Intelligence, Changzhou University, Changzhou, China
3 shared publications Saqib Mamoon
School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China
2 shared publicationsTao Zhou
School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China
2 shared publications
Latest Publications
- Open Access
- 23 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Systems and Rehabilitation Engineering
- Vol. 34 (15344320) , 1826-1836
Causal-Augmented Source-Free Domain Adaptation With Scale-Free Transformer for Schizophrenia Classification
Brain functional networks (BFNs) derived from multi-site fMRI data have been widely explored using transformer-based models to extract discriminative connectivity features for the diagnosis of psychiatric disorders, such as schizophrenia (SZ). However, existing transformer-based methods ignore physiological priors like the scale-free property, causing the attention to fail to capture the intrinsic topology of BFNs. Additionally, multisite heterogeneity makes source-free domain adaptation (DA) essential in clinical practice where data sharing is restricted. However, existing methods mainly rely on heuristic data augmentations or pseudo-labeling, without leveraging the intrinsic inter-regional dependencies, resulting in poor robustness to cross-site variability. To overcome these challenges, we proposed a source-free DA framework based on a scale-free transformer encoder for SZ classification. The transformer encoder was pre-trained on labeled source domains to capture discriminative connectivity patterns while integrating a scale-free prior to bias the attention toward hub nodes. The pretrained encoder and classifier were used to initialize the target model, where the encoder learned latent representations derived from inter-regional interactions for causal graph construction. To enhance robustness, the causal structure was perturbed via random permutation and counterfactual interventions, while entropy minimization jointly optimized the encoder and predictor to learn domain-invariant representations. Results showed that our method outperformed the other 4 transformer, 7 DA, 6 multi-site and 6 state-of-the-art methods across two SZ datasets (87.18%±0.91% and 88.39%±0.13%). Ablation results highlighted the contributions of the causal, permutation, counterfactual, and entropy minimization constraints to the performance improvement. Furthermore, the identified discriminative temporal regions provided insights into the dysfunctional neural-mechanisms in SZ.
- 23 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Systems and Rehabilitation Engineering
- Vol. 34 (15344320) , 1826-1836
Open AccessCausal-Augmented Source-Free Domain Adaptation With Scale-Free Transformer for Schizophrenia Classification
Brain functional networks (BFNs) derived from multi-site fMRI data have been widely explored using transformer-based models to extract discriminative connectivity features for the diagnosis of psychiatric disorders, such as schizophrenia (SZ). However, existing transformer-based methods ignore physiological priors like the scale-free property, causing the attention to fail to capture the intrinsic topology of BFNs. Additionally, multisite heterogeneity makes source-free domain adaptation (DA) essential in clinical practice where data sharing is restricted. However, existing methods mainly rely on heuristic data augmentations or pseudo-labeling, without leveraging the intrinsic inter-regional dependencies, resulting in poor robustness to cross-site variability. To overcome these challenges, we proposed a source-free DA framework based on a scale-free transformer encoder for SZ classification. The transformer encoder was pre-trained on labeled source domains to capture discriminative connectivity patterns while integrating a scale-free prior to bias the attention toward hub nodes. The pretrained encoder and classifier were used to initialize the target model, where the encoder learned latent representations derived from inter-regional interactions for causal graph construction. To enhance robustness, the causal structure was perturbed via random permutation and counterfactual interventions, while entropy minimization jointly optimized the encoder and predictor to learn domain-invariant representations. Results showed that our method outperformed the other 4 transformer, 7 DA, 6 multi-site and 6 state-of-the-art methods across two SZ datasets (87.18%±0.91% and 88.39%±0.13%). Ablation results highlighted the contributions of the causal, permutation, counterfactual, and entropy minimization constraints to the performance improvement. Furthermore, the identified discriminative temporal regions provided insights into the dysfunctional neural-mechanisms in SZ.
- Open Access
- 22 February 2026
- Published by Taylor & Francis in Connection Science
- Vol. 38 (1)
Feature interaction modelling using a similarity-based adaptive graph attention network for click-through rate prediction
- 22 February 2026
- Published by Taylor & Francis in Connection Science
- Vol. 38 (1)
Open AccessFeature interaction modelling using a similarity-based adaptive graph attention network for click-through rate prediction
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Systems and Rehabilitation Engineering
- Vol. 34, 4371-4383
Time-Delay-Aware Causal Autoencoder for Neurological Disorder Recognition
Brain networks have become a crucial approach for characterizing interactions among brain regions, offering a powerful tool for understanding the pathological mechanisms underlying neurological disorders. In recent years, numerous brain network estimation methods have been developed based on correlation-based metrics. However, these approaches can only reflect statistical dependencies between brain regions. To address this limitation, various methods have been proposed to estimate causal effects, i.e., the direction of information flow, among brain regions. Unfortunately, many existing approaches assume uniform information transmission delays across different brain regions, which can introduce estimation bias and ultimately reduce the accuracy of neurological disorder identification. To address this issue, we propose a Time-Delay-Aware Causal Autoencoder (TDCA) that jointly estimates causal effects and transmission delays across brain regions. Furthermore, we introduce two constraints to ensure that the constructed brain network complies with established neurophysiological principles. Experimental results on two publicly datasets demonstrate the effectiveness and superiority of the proposed method.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Systems and Rehabilitation Engineering
- Vol. 34, 4371-4383
Open AccessTime-Delay-Aware Causal Autoencoder for Neurological Disorder Recognition
Brain networks have become a crucial approach for characterizing interactions among brain regions, offering a powerful tool for understanding the pathological mechanisms underlying neurological disorders. In recent years, numerous brain network estimation methods have been developed based on correlation-based metrics. However, these approaches can only reflect statistical dependencies between brain regions. To address this limitation, various methods have been proposed to estimate causal effects, i.e., the direction of information flow, among brain regions. Unfortunately, many existing approaches assume uniform information transmission delays across different brain regions, which can introduce estimation bias and ultimately reduce the accuracy of neurological disorder identification. To address this issue, we propose a Time-Delay-Aware Causal Autoencoder (TDCA) that jointly estimates causal effects and transmission delays across brain regions. Furthermore, we introduce two constraints to ensure that the constructed brain network complies with established neurophysiological principles. Experimental results on two publicly datasets demonstrate the effectiveness and superiority of the proposed method.
- 6 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 30 (4) , 2986-2999
Joint Dynamic Brain Network Estimation and Graph Representation Learning for the Recognition of Neurological Disorders
Recently, Graph Neural Networks (GNNs) have shown significant improvements in the recognition of neurological disorders by incorporating brain networks/graphs. However, most existing approaches have three main limitations. First, these methodologies rely on precomputed brain networks as input, typically derived from statistical metrics (e.g., Pearson correlation), which are inherently not learnable. Second, methods often assume that the magnitude of the brain interactions remains constant across the whole scan duration. Third, representations produced by models often lack interpretability and robustness when applied across brain disorders. To address these limitations, we propose a novel model called the Effective Brain Inference Graph Neural Network (EBIGNN), which infers dynamic Effective Connectivity (dEC) to characterize brain networks and is trained with direct feedback from downstream tasks within a unified end-to-end framework. EBIGNN is highly flexible in learning the most relevant graph structures customized to the specific underlying brain condition. The proposed model offers strong interpretability, providing valuable insights into the temporal evolution and altered connectivity patterns essential for understanding brain disorders. The model is validated on three publicly available datasets, demonstrating superior performance compared to other state-of-the-art methods. Moreover, the findings are consistent with previous neuroimaging-derived evidence of biomarkers, underscoring the model's robustness in clinical settings. The code will be released at https://github.com/smbiomedical/EBIGNN.
- 6 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 30 (4) , 2986-2999
Joint Dynamic Brain Network Estimation and Graph Representation Learning for the Recognition of Neurological Disorders
Recently, Graph Neural Networks (GNNs) have shown significant improvements in the recognition of neurological disorders by incorporating brain networks/graphs. However, most existing approaches have three main limitations. First, these methodologies rely on precomputed brain networks as input, typically derived from statistical metrics (e.g., Pearson correlation), which are inherently not learnable. Second, methods often assume that the magnitude of the brain interactions remains constant across the whole scan duration. Third, representations produced by models often lack interpretability and robustness when applied across brain disorders. To address these limitations, we propose a novel model called the Effective Brain Inference Graph Neural Network (EBIGNN), which infers dynamic Effective Connectivity (dEC) to characterize brain networks and is trained with direct feedback from downstream tasks within a unified end-to-end framework. EBIGNN is highly flexible in learning the most relevant graph structures customized to the specific underlying brain condition. The proposed model offers strong interpretability, providing valuable insights into the temporal evolution and altered connectivity patterns essential for understanding brain disorders. The model is validated on three publicly available datasets, demonstrating superior performance compared to other state-of-the-art methods. Moreover, the findings are consistent with previous neuroimaging-derived evidence of biomarkers, underscoring the model's robustness in clinical settings. The code will be released at https://github.com/smbiomedical/EBIGNN.
- Open Access
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Systems and Rehabilitation Engineering
- Vol. 33, 2770-2780
Brain-Shapelet: A Framework for Capturing Instantaneous Abnormalities in Brain Activity for Autism Spectrum Disorder Diagnosis
Some symptoms of Autism Spectrum Disorder (ASD), such as anxiety and depression, often manifest intermittently rather than continuously, complicating the identification of reliable pathophysiological biomarkers. Meanwhile, functional connectivity networks (FCNs) generate high-dimensional connectomes, making it difficult to accurately capture instantaneous abnormal biomarkers of neurological disorders. To address this issue, we propose a framework, called Brain-Shapelet, to extract discriminative subsequences (Shapelets) from functional magnetic resonance imaging (fMRI) data for capturing instantaneous abnormalities in brain activity. It applies random walk algorithm on group-representative brain network to obtain brain region sets, and aggregates their blood oxygen level-dependent (BOLD) signals to extract Shapelets that reflect the associations between different brain regions at the same time point. Specially, we develop a feature selection strategy to reduce redundancy in Shapelets and optimize classification performance. Brain-Shapelet places greater emphasis on short-term brain activity alterations, allowing it to capture instantaneous abnormalities more effectively. It is evaluated on the ABIDE dataset and achieves a classification accuracy of 82.8%, significantly outperforming traditional brain network modeling methods. The proposed co-occurrence rate, occurrence frequency, and Gini coefficient metrics quantify the contributions of brain regions from the perspective of Shapelets, offering valuable insights for ASD diagnosis.
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Systems and Rehabilitation Engineering
- Vol. 33, 2770-2780
Open AccessBrain-Shapelet: A Framework for Capturing Instantaneous Abnormalities in Brain Activity for Autism Spectrum Disorder Diagnosis
Some symptoms of Autism Spectrum Disorder (ASD), such as anxiety and depression, often manifest intermittently rather than continuously, complicating the identification of reliable pathophysiological biomarkers. Meanwhile, functional connectivity networks (FCNs) generate high-dimensional connectomes, making it difficult to accurately capture instantaneous abnormal biomarkers of neurological disorders. To address this issue, we propose a framework, called Brain-Shapelet, to extract discriminative subsequences (Shapelets) from functional magnetic resonance imaging (fMRI) data for capturing instantaneous abnormalities in brain activity. It applies random walk algorithm on group-representative brain network to obtain brain region sets, and aggregates their blood oxygen level-dependent (BOLD) signals to extract Shapelets that reflect the associations between different brain regions at the same time point. Specially, we develop a feature selection strategy to reduce redundancy in Shapelets and optimize classification performance. Brain-Shapelet places greater emphasis on short-term brain activity alterations, allowing it to capture instantaneous abnormalities more effectively. It is evaluated on the ABIDE dataset and achieves a classification accuracy of 82.8%, significantly outperforming traditional brain network modeling methods. The proposed co-occurrence rate, occurrence frequency, and Gini coefficient metrics quantify the contributions of brain regions from the perspective of Shapelets, offering valuable insights for ASD diagnosis.
- Open Access
- 1 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Systems and Rehabilitation Engineering
- Vol. 32 (15344320) , 3729-3740
UCLN: Toward the Causal Understanding of Brain Disorders With Temporal Lag Dynamics
Resting-state functional magnetic resonance imaging (rs-fMRI) has emerged as a powerful tool for exploring interactions among brain regions. A growing body of research is actively investigating various computational approaches for estimating causal effects among brain regions. Compared to traditional methods, causal relationship reveals the causal influences among distinct brain regions, offering a deeper understanding of brain network dynamics. However, existing methods either neglect the concept of temporal lag across brain regions or set the temporal lag value to a fixed value. To address this limitation, we propose a Unified Causal and Temporal Lag Network (termed UCLN) that jointly learns the causal effects and temporal lag values among brain regions. Our method effectively captures variations in temporal lag between distant brain regions by avoiding the predefined lag value across the entire brain. The brain networks obtained are directed and weighted graphs, enabling a more comprehensive disentanglement of complex interactions. In addition, we also introduce three guiding mechanisms for efficient brain network modeling. The proposed method outperforms state-of-the-art approaches in classification accuracy on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. Our findings indicate that the method not only achieves superior classification but also successfully identifies crucial neuroimaging biomarkers associated with the disease.
- 1 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Systems and Rehabilitation Engineering
- Vol. 32 (15344320) , 3729-3740
Open AccessUCLN: Toward the Causal Understanding of Brain Disorders With Temporal Lag Dynamics
Resting-state functional magnetic resonance imaging (rs-fMRI) has emerged as a powerful tool for exploring interactions among brain regions. A growing body of research is actively investigating various computational approaches for estimating causal effects among brain regions. Compared to traditional methods, causal relationship reveals the causal influences among distinct brain regions, offering a deeper understanding of brain network dynamics. However, existing methods either neglect the concept of temporal lag across brain regions or set the temporal lag value to a fixed value. To address this limitation, we propose a Unified Causal and Temporal Lag Network (termed UCLN) that jointly learns the causal effects and temporal lag values among brain regions. Our method effectively captures variations in temporal lag between distant brain regions by avoiding the predefined lag value across the entire brain. The brain networks obtained are directed and weighted graphs, enabling a more comprehensive disentanglement of complex interactions. In addition, we also introduce three guiding mechanisms for efficient brain network modeling. The proposed method outperforms state-of-the-art approaches in classification accuracy on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. Our findings indicate that the method not only achieves superior classification but also successfully identifies crucial neuroimaging biomarkers associated with the disease.
- Open Access
- 20 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Systems and Rehabilitation Engineering
- Vol. 32 (15344320) , 3084-3094
Learnable Brain Connectivity Structures for Identifying Neurological Disorders
Brain networks/graphs have been widely recognized as powerful and efficient tools for identifying neurological disorders. In recent years, various graph neural network models have been developed to automatically extract features from brain networks. However, a key limitation of these models is that the inputs, namely brain networks/graphs, are constructed using predefined statistical metrics (e.g., Pearson correlation) and are not learnable. The lack of learnability restricts the flexibility of these approaches. While statistically-specific brain networks can be highly effective in recognizing certain diseases, their performance may not exhibit robustness when applied to other types of brain disorders. To address this issue, we propose a novel module called Brain Structure Inference (termed BSI), which can be seamlessly integrated with multiple downstream tasks within a unified framework, enabling end-to-end training. It is highly flexible to learn the most beneficial underlying graph structures directly for specific downstream tasks. The proposed method achieves classification accuracies of 74.83% and 79.18% on two publicly available datasets, respectively. This suggests an improvement of at least 3% over the best-performing existing methods for both tasks. In addition to its excellent performance, the proposed method is highly interpretable, and the results are generally consistent with previous findings.
- 20 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Systems and Rehabilitation Engineering
- Vol. 32 (15344320) , 3084-3094
Open AccessLearnable Brain Connectivity Structures for Identifying Neurological Disorders
Brain networks/graphs have been widely recognized as powerful and efficient tools for identifying neurological disorders. In recent years, various graph neural network models have been developed to automatically extract features from brain networks. However, a key limitation of these models is that the inputs, namely brain networks/graphs, are constructed using predefined statistical metrics (e.g., Pearson correlation) and are not learnable. The lack of learnability restricts the flexibility of these approaches. While statistically-specific brain networks can be highly effective in recognizing certain diseases, their performance may not exhibit robustness when applied to other types of brain disorders. To address this issue, we propose a novel module called Brain Structure Inference (termed BSI), which can be seamlessly integrated with multiple downstream tasks within a unified framework, enabling end-to-end training. It is highly flexible to learn the most beneficial underlying graph structures directly for specific downstream tasks. The proposed method achieves classification accuracies of 74.83% and 79.18% on two publicly available datasets, respectively. This suggests an improvement of at least 3% over the best-performing existing methods for both tasks. In addition to its excellent performance, the proposed method is highly interpretable, and the results are generally consistent with previous findings.
- 6 March 2024
- Published by Elsevier in Medical Image Analysis
- Vol. 94, 103133
Inferring brain causal and temporal-lag networks for recognizing abnormal patterns of dementia
- 6 March 2024
- Published by Elsevier in Medical Image Analysis
- Vol. 94, 103133
Inferring brain causal and temporal-lag networks for recognizing abnormal patterns of dementia
- Open Access
- 19 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Systems and Rehabilitation Engineering
- Vol. 32 (15344320) , 112-121
Deep Causality Variational Autoencoder Network for Identifying the Potential Biomarkers of Brain Disorders
Identifying causality from observational time-series data is a key problem in dealing with complex dynamic systems. Inferring the direction of connection between brain regions (i.e., causality) has become the central topic in the domain of fMRI. The purpose of this study is to obtain causal graphs that characterize the causal relationship between brain regions based on time series data. To address this issue, we designed a novel model named deep causal variational autoencoder (CVAE) to estimate the causal relationship between brain regions. This network contains a causal layer that can estimate the causal relationship between different brain regions directly. Compared with previous approaches, our method relaxes many constraints on the structure of underlying causal graph. Our proposed method achieves excellent performance on both the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and the Autism Brain Imaging Data Exchange 1 (ABIDE1) databases. Moreover, the experimental results show that deep CVAE has promising applications in the field of brain disease identification.
- 19 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Systems and Rehabilitation Engineering
- Vol. 32 (15344320) , 112-121
Open AccessDeep Causality Variational Autoencoder Network for Identifying the Potential Biomarkers of Brain Disorders
Identifying causality from observational time-series data is a key problem in dealing with complex dynamic systems. Inferring the direction of connection between brain regions (i.e., causality) has become the central topic in the domain of fMRI. The purpose of this study is to obtain causal graphs that characterize the causal relationship between brain regions based on time series data. To address this issue, we designed a novel model named deep causal variational autoencoder (CVAE) to estimate the causal relationship between brain regions. This network contains a causal layer that can estimate the causal relationship between different brain regions directly. Compared with previous approaches, our method relaxes many constraints on the structure of underlying causal graph. Our proposed method achieves excellent performance on both the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and the Autism Brain Imaging Data Exchange 1 (ABIDE1) databases. Moreover, the experimental results show that deep CVAE has promising applications in the field of brain disease identification.
- 7 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 27 (6) , 2990-3001
A Structure-Guided Effective and Temporal-Lag Connectivity Network for Revealing Brain Disorder Mechanisms
Brain network provides important insights for the diagnosis of many brain disorders, and how to effectively model the brain structure has become one of the core issues in the domain of brain imaging analysis. Recently, various computational methods have been proposed to estimate the causal relationship (i.e., effective connectivity) between brain regions. Compared with traditional correlation-based methods, effective connectivity can provide the direction of information flow, which may provide additional information for the diagnosis of brain diseases. However, existing methods either ignore the fact that there is a temporal-lag in the information transmission across brain regions, or simply set the temporal-lag value between all brain regions to a fixed value. To overcome these issues, we design an effective temporal-lag neural network (termed ETLN) to simultaneously infer the causal relationships and the temporal-lag values between brain regions, which can be trained in an end-to-end manner. In addition, we also introduce three mechanisms to better guide the modeling of brain networks. The evaluation results on the Alzheimer's Disease Neuroimaging Initiative (ADNI) database demonstrate the effectiveness of the proposed method.
- 7 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 27 (6) , 2990-3001
A Structure-Guided Effective and Temporal-Lag Connectivity Network for Revealing Brain Disorder Mechanisms
Brain network provides important insights for the diagnosis of many brain disorders, and how to effectively model the brain structure has become one of the core issues in the domain of brain imaging analysis. Recently, various computational methods have been proposed to estimate the causal relationship (i.e., effective connectivity) between brain regions. Compared with traditional correlation-based methods, effective connectivity can provide the direction of information flow, which may provide additional information for the diagnosis of brain diseases. However, existing methods either ignore the fact that there is a temporal-lag in the information transmission across brain regions, or simply set the temporal-lag value between all brain regions to a fixed value. To overcome these issues, we design an effective temporal-lag neural network (termed ETLN) to simultaneously infer the causal relationships and the temporal-lag values between brain regions, which can be trained in an end-to-end manner. In addition, we also introduce three mechanisms to better guide the modeling of brain networks. The evaluation results on the Alzheimer's Disease Neuroimaging Initiative (ADNI) database demonstrate the effectiveness of the proposed method.