Article
Computer Science, Artificial Intelligence
Jiawen Wei, Zhifeng Qiu, Fangyuan Wang, Wenwei Lin, Ning Gui, Weihua Gui
Summary: The article proposes a novel online feature selection framework that quantitatively identifies the importance of features for control by analyzing the response of deep reinforcement learning algorithms in both the real world and a virtual peer. This framework achieves better or comparable feature combinations to those provided by human experts and recent feature selection baselines.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Shiye Lei, Fengxiang He, Yancheng Yuan, Dacheng Tao
Summary: This article finds that neural networks with less variability in decision boundaries have better generalizability. The experiments show significant negative correlations between decision boundary variability and generalizability. The article introduces the concepts of algorithm DB variability and (epsilon, eta)-data DB variability to measure variability in decision boundaries.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Green & Sustainable Science & Technology
Lei Shi, Jia Luo, Peiying Zhang, Hongqi Han, Didier El Baz, Gang Cheng, Zeyu Liang
Summary: The study focuses on the check-in behaviors of users in location-based social networks in urban living, proposing a novel method named STAT built on Transformer architecture to learn user preferences through graph deep learning. Extensive experiments demonstrate the superiority of STAT over state-of-the-art POI recommendation methods.
Article
Computer Science, Artificial Intelligence
Xinyi Xu, Zhengyang Wang, Cheng Deng, Hao Yuan, Shuiwang Ji
Summary: This study proposes an improved and interpretable grouping method to enhance the performance and interpretability of deep metric learning. The method utilizes attention mechanism with learnable queries to capture group-specific information. The results demonstrate that the proposed method consistently outperforms prior methods across various evaluation metrics, datasets, base models, and loss functions.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Biochemical Research Methods
Boqiao Lai, Jinbo Xu
Summary: Experimental protein function annotation cannot keep up with the rapid expansion of sequence databases. Computational methods for predicting protein function are fast but not very accurate. Researchers have developed a method, called GAT-GO, that improves protein function prediction by combining predicted protein structure and sequence information. The experimental results show that GAT-GO outperforms the latest deep learning methods and homology-based methods.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Computer Science, Information Systems
Yuantao Chen, Runlong Xia, Kai Yang, Ke Zou
Summary: To address the limitations of existing image inpainting methods, such as lack of authenticity, inflexible handling of missing and non-missing regions, and ineffective treatment of image feature information, we propose an image restoration method that combines Semantic Priors and Deep Attention Residual Group. Our method involves the use of a Semantic Priors Network to learn semantic prior information and complete the missing regions, a Deep Attention Residual Group to focus on the missing regions and learn adaptive channel features, and a Full-scale Skip Connection to repair the missing regions by combining low-level and high-level feature maps. Experimental results on CelebA-HQ and Paris StreetView datasets demonstrate the superiority of our method over current state-of-the-art image restoration methods.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2023)
Article
Mathematics
Diana-Laura Borza, Adrian Sergiu Darabant, Tudor Alexandru Ileni, Alexandru-Ion Marinescu
Summary: This paper introduces an online knowledge distillation framework that combines the predictions of lightweight networks into a powerful ensemble using an attention mechanism. Experimental results show that this framework improves the accuracy of knowledge-distilled students and in some cases outperforms deeper individually trained models.
Article
Biochemical Research Methods
Yahui Long, Jiawei Luo, Yu Zhang, Yan Xia
Summary: This study proposes a novel deep learning framework, GATMDA, for predicting human microbe-disease associations by fully exploiting multiple sources of biomedical data. Comprehensive experiments on two datasets demonstrated that the proposed model consistently outperformed baseline methods. Case studies further confirmed the effectiveness of the GATMDA model.
BRIEFINGS IN BIOINFORMATICS
(2021)
Article
Computer Science, Artificial Intelligence
Xuan Wang, Zhigang Zhu
Summary: Contextual information is important in computer vision tasks to improve accuracy, and it includes appearance context information, semantic context information, and contextual information not present in the image itself. This survey reviews the different types and levels of context information used in computer vision tasks and discusses available machine learning models and datasets. It also compares context-based and context-free integration in image-based and video-based tasks and proposes future directions for context learning and utilization.
COMPUTER VISION AND IMAGE UNDERSTANDING
(2023)
Article
Multidisciplinary Sciences
Jin-Hong Du, Zhanrui Cai, Kathryn Roeder
Summary: In this study, the researchers propose a probabilistic variational autoencoder model, scVAEIT, for integrating and imputing multimodal datasets with mosaic measurements. The model effectively combines different panels of measurements and accurately imputes missing molecular layers. Validation results show that scVAEIT robustly imputes missing modalities and features of cells biologically different from the training data, and it adjusts for batch effects while maintaining biological variation. The study demonstrates that scVAEIT significantly improves integration and imputation across unseen cell types, different technologies, and different tissues.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2022)
Article
Computer Science, Interdisciplinary Applications
Xuming An, Pengchang Li, Chen Zhang
Summary: In this study, a novel deep cascade-learning model is proposed to address two tasks in immunofixation electrophoresis (IFE) diagnosis: determining the existence of M-protein and identifying its isotype. The model combines a positive-negative classifier and an isotype classifier, and incorporates an attention mechanism and domain knowledge. The model outperforms existing methods in terms of recognition evaluation metrics and effectively captures the co-location of dense bands in different lanes.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2023)
Article
Computer Science, Artificial Intelligence
Yuantao Chen, Runlong Xia, Kai Yang, Ke Zou
Summary: This paper proposes an improved image inpainting network using a multi-scale feature module and improved attention module. The network addresses issues in deep learning-based image inpainting algorithms, such as information loss in deep level features and the neglect of semantic features. The proposed network generates better inpainting results by reducing information loss and enhancing the ability to restore texture and semantic features.
COMPUTER VISION AND IMAGE UNDERSTANDING
(2024)
Article
Chemistry, Analytical
Yi Li, Lingna Wang, Zeji Wang
Summary: In this study, a feature-enhancement- and channel-attention-guided single-shot detector (FCSSD) was proposed to improve object detection performance. By utilizing four modules, contextual and semantic information were explored, multi-scale features were refined, and channel weights were balanced, resulting in excellent detection performance for multi-scale object detection.
Article
Computer Science, Artificial Intelligence
Jiazhong Chen, Qingqing Li, Hefei Ling, Dakai Ren, Ping Duan
Summary: This paper introduces a multisensory framework for video saliency prediction using audio and visual signals, which includes four modules to be implemented through a deep learning network architecture. The method shows a significant improvement over existing saliency models that do not take into account audio stimuli, according to numerical and visual results.
Article
Computer Science, Information Systems
Dakai Ren, Jiazhong Chen, Jian Zhong, Zhaoming Lu, Tao Jia, Zongyi Li
Summary: An innovative bilinear pooling-based attention mechanism is introduced to extract second-order contextual information. To enhance the robustness of gaze-related features against spatial misalignment, an attention-in-attention method is proposed. The method shows outstanding performance in gaze estimation and surpasses the current state-of-the-art by a large margin.
JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION
(2021)
Article
Public, Environmental & Occupational Health
Rui Li, Yan Li, Zhuoru Zou, Yiming Liu, Xinghui Li, Guihua Zhuang, Mingwang Shen, Lei Zhang
Summary: This study evaluated the impact of vaccine scale-up and potential reduction in vaccine effectiveness on the COVID-19 epidemic and social restoration in the United States. The results showed that current COVID-19 vaccines remain effective against the SARS-CoV-2 variant, and a vaccination coverage of 70% would be sufficient to restore social activities to a pre-pandemic level. However, a reduction in vaccine effectiveness could result in a surge of the epidemic. Therefore, multiple measures should be implemented to counter the challenges posed by new SARS-CoV-2 variants.
FRONTIERS IN PUBLIC HEALTH
(2022)
Article
Biology
Tobias Moll, Valerie Odon, Calum Harvey, Mark O. Collins, Andrew Peden, John Franklin, Emily Graves, Jack N. G. Marshall, Cleide dos Santos Souza, Sai Zhang, Lydia Castelli, Guillaume Hautbergue, Mimoun Azzouz, David Gordon, Nevan Krogan, Laura Ferraiuolo, Michael P. Snyder, Pamela J. Shaw, Jan Rehwinkel, Johnathan Cooper-Knock
Summary: This study identifies a link between reduced expression of EXOSC2 and reduced SARS-CoV-2 replication. Increased expression of EXOSC2 is associated with higher risk of clinical COVID-19. The study also reveals interaction between the SARS-CoV-2 RNA polymerase and most of the human RNA exosome components.
LIFE SCIENCE ALLIANCE
(2023)
Review
Geriatrics & Gerontology
Jun Zhang, Yushan Yu, Mirko Petrovic, Xiaomei Pei, Qing-Bao Tian, Lei Zhang, Wei-Hong Zhang
Summary: Based on our research on LTCFs, we found that the impacts of the COVID-19 pandemic on LTCFs include infection rates, hospitalization rates, case fatality rates, and mortality rates. Cancelling visits, restricting new admissions, limiting communal dining and group activities, and increasing vaccination can significantly reduce the infection rates among residents and staff.
Article
Immunology
Jason J. Ong, Ross D. Booton, Joseph D. Tucker, Weiming Tang, Peter Vickerman, Lei Zhang, Kate M. Mitchell
Summary: Crowdsourced interventions for enhancing HIV self-testing among MSM in China were found to be highly cost-effective, with one-off interventions achieving cost savings in certain cities such as Guangzhou and Qingdao. Further research is needed to assess the scalability of crowdsourced HIV prevention interventions in other settings and populations.
Article
Environmental Sciences
Yuming Guo, Yao Wu, Tingting Ye, Lei Zhang, Amanda Johnson, Shanshan Li
Summary: This study conducted a national level analysis in Brazil from 2000 to 2015 to accurately estimate the causal relationship between PM2.5 exposure and hospitalisations. A novel approach using a panel analysis was employed to control for seasonality and long-term trends. The findings showed that a 10 μg/m3 increase in PM2.5 concentrations was associated with a 1.06% increase in hospitalisations, with larger effects observed among children and the elderly.
ENVIRONMENT INTERNATIONAL
(2023)
Review
Infectious Diseases
Cham-mill Kim, Victor Zhao, Maeve Brito De Mello, Rachel Baggaley, Cheryl C. Johnson, Erica Spielman, Christopher K. Fairley, Lei Zhang, Henry de Vries, Jeffrey Klausner, Rui Zhao, Jason J. Ong
Summary: There is no evidence for optimal frequency of STI screening for people using pre-exposure prophylaxis for HIV, although more frequent screening could reduce delayed diagnoses and incidence.
INTERNATIONAL JOURNAL OF INFECTIOUS DISEASES
(2023)
Article
Public, Environmental & Occupational Health
Xianglong Xu, Eric P. F. Chow, Christopher K. Fairley, Marcus Chen, Ivette Aguirre, Jane Goller, Jane Hocking, Natalie Carvalho, Lei Zhang, Jason J. Ong
Summary: This study aimed to identify determinants and predict re-testing and re-infection within 1 year among heterosexuals with chlamydia, in order to identify potential candidates for patient-delivered partner therapy (PDPT). The results showed a low re-testing rate and high re-infection rate. Further interventions are needed to improve re-testing rates, reduce re-infection rates, and better target individuals suitable for PDPT.
FRONTIERS IN PUBLIC HEALTH
(2023)
Article
Infectious Diseases
Jason J. Ong, Aaron Lim, Catriona Bradshaw, David Taylor-Robinson, Magnus Unemo, Paddy J. Horner, Peter Vickerman, Lei Zhang
Summary: This study aims to evaluate the cost-effectiveness of different testing strategies for Mycoplasma genitalium (MG) in men who have sex with men (MSM). The results showed that testing only symptomatic MSM is the most cost-effective approach.
SEXUALLY TRANSMITTED INFECTIONS
(2023)
Article
Health Care Sciences & Services
Rui Li, Mingwang Shen, Hanting Liu, Lu Bai, Lei Zhang
Summary: This article proposes a theoretical framework for a community-based early warning system that detects temperature abnormalities in the community using a collective network of infrared thermometer-enabled smartphone devices. The framework utilizes advanced AI technology on cloud computing platforms to identify outbreaks in a timely manner through the detection of geospatial temperature abnormalities based on mass data collection, cloud-based computing and analysis, decision-making, and feedback. While the initial model training process is relatively long, the system may be feasible based on its public acceptance, technical practicality, and value for money.
JMIR FORMATIVE RESEARCH
(2023)
Article
Health Care Sciences & Services
Lei Zhang, Hanting Liu, Zhuoru Zou, Shu Su, Jason J. Ong, Fanpu Ji, Fuqiang Cui, Po -lin Chan, Qin Ning, Rui Li, Mingwang Shen, Christopher K. Fairley, Lan Liu, Wai-Kay Seto, William C. W. Wong
Summary: Shared-care models with HBV testing, follow up and referring of predetermined conditions to specialty care at an appropriate time, especially antiviral treatment initiation in primary care, are highly effective and cost-effective in China.
LANCET REGIONAL HEALTH-WESTERN PACIFIC
(2023)
Article
Health Care Sciences & Services
Jinli Liu, Min Liu, Zhonglin Chai, Chao Li, Yanan Wang, Mingwang Shen, Guihua Zhuang, Lei Zhang
Summary: This study predicts the disease burden and economic burden trends of diabetes in China from 2020 to 2030, and explores the spatial disparities. Bayesian modeling is used to estimate the prevalence and disability-adjusted life-year rates over time. The results show an increase in diabetes prevalence and economic burden, with strong spatial aggregation in northern Chinese regions. The economic burden of diabetes is projected to grow faster than the country's GDP.
LANCET REGIONAL HEALTH-WESTERN PACIFIC
(2023)
Editorial Material
Medicine, General & Internal
Zhuoru Zou, Lei Zhang
Article
Public, Environmental & Occupational Health
Jing Wang, Patrick Kwan, Gong Zhang, Mingwang Shen, Loretta Piccenna, Terence J. O'Brien, Lei Zhang
Summary: China is faced with a growing aging population and understanding the health status of older adults is crucial for resource allocation and healthcare provision. This study aimed to comprehensively assess the disability level and identify risk factors associated with disability among older adults in China. A multidimensional ability assessment survey was used to evaluate daily living activities, mental status, sensory and communication abilities, and social participation. Demographic risk factors were analyzed using logistic regression and the correlations between the four dimensions of ability were assessed.
JMIR PUBLIC HEALTH AND SURVEILLANCE
(2023)
Article
Gastroenterology & Hepatology
Rui Li, Mingwang Shen, Jason J. Ong, Fuqiang Cui, Wenyi Hu, Polin Chan, Zhuoru Zou, Shu Su, Hangting Liu, Lei Zhang, Wai-Kay Seto, William C. W. Wong
Summary: This study explores the key developments and optimal intervention strategies needed to achieve WHO hepatitis B elimination targets by 2030 in China. It highlights that China can realize the HBV elimination targets in the incidence by 2025, and by upscaling diagnostic, linkage-to-care, and treatment coverages, up to 2 million lives could potentially be saved from HBV-related deaths.
Article
Oncology
Tianhao Shan, Xianhui Ran, Huizhang Li, Guoshuang Feng, Siwei Zhang, Xuehong Zhang, Lei Zhang, Lingeng Lu, Lan An, Ruiying Fu, Kexin Sun, Shaoming Wang, Ru Chen, Li Li, Wanqing Chen, Wenqiang Wei, Hongmei Zeng, Jie He
Summary: This study reveals disparities in the stage of liver cancer diagnosis among different populations in China, with higher proportions of advanced stage (III-IV) cases in women, patients aged 60 and above, drinkers, and those without a family history of cancer. Furthermore, the stage distribution in China is significantly higher than that in the USA.
JOURNAL OF THE NATIONAL CANCER CENTER
(2023)