Sheng Wang

Sheng Wang

Auto-generated profile, based on publicly available information

Sheng Wang

Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 March 2026 in Nature Biomedical Engineering
Verified
Journal Article
Closed access
DBDechao Bu
Dechao Bu
JSJingbo Sun
Jingbo Sun
KLKun Li
Kun Li
ZHZihao He
Zihao He
WHWei Huang
Wei Huang
YZYi Zhao
Yi Zhao

13 October 2025 in Nature Communications
Verified
Journal Article
Open access
SWSheng Wang
Sheng Wang
FZFangyuan Zhao
Fangyuan Zhao
DBDechao Bu
Dechao Bu
YLYunwei Lu
Yunwei Lu
MGMing Gong
Ming Gong
YZYi Zhao
Yi Zhao
Large language models can lighten the workload of clinicians and patients, yet their responses often include fabricated evidence, outdated knowledge, and insufficient medical specificity. We introduce a general retrieval-augmented question-answering framework that continuously gathers up-to-date, high-quality medical knowledge and generates evidence-traceable responses. Here we show that this approach significantly improves the evidence validity, medical expertise, and timeliness of large language model outputs, thereby enhancing their overall quality and credibility. Evaluation against 15,530 objective questions, together with two physician-curated clinical test sets covering evidence-based medical practice and medical order explanation, confirms the improvements. In blinded trials, resident physicians indicate meaningful assistance in 87.00% of evidence-based medical scenarios, and lay users find it helpful in 90.09% of medical order explanations. These findings demonstrate a practical route to trustworthy, general-purpose language assistants for clinical applications.

1 April 2026 in Communications Biology
Verified
Journal Article
Open access
PDF file
FZFangyuan Zhao
Fangyuan Zhao
SWSheng Wang
Sheng Wang
ZWZhikang Wang
Zhikang Wang
YCYan Cui
Yan Cui
YZYi Zhao
Yi Zhao
ZYZhiyuan Yuan
Zhiyuan Yuan
Mining multi-cellular spatial patterns associated with biological events from high-resolution spatial omics data remains a fundamental challenge. While current computational methods have advanced from pairwise associations to identifying higher-order spatial domains, they often lack the granularity to resolve subtle local architectural shifts or the statistical framework to quantify condition-specificity. Here, we present FDPMining (Frequent and Distinctive spatial Patterns Mining), a computational framework that reformulates the biological problem of pattern discovery into a scalable data mining task through a Neighborhood-to-Sequence (N2S) encoding strategy. This transformation uniquely converts spatial grid neighborhoods for each cell into lossless and reversible numerical sequences, enabling efficient and scalable discovery of FDPs (Frequent and Distinctive spatial Patterns) via data mining algorithms. Our approach systematically explores the vast combinatorial space of cellular arrangements to identify FDPs associated with specific biological conditions. To enable spatial traceability, we further develop FDPs-Mapping, a spatial reconstruction component that maps identified patterns back to their original tissue context. This advancement allows researchers to examine and interpret patterns directly in situ. In extensive benchmarking, FDPMining demonstrates superior sensitivity in capturing subtle and condition-specific differences, outperforming state-of-the-art pairwise and higher-order pattern discovery methods. We applied our framework across diverse biological systems and spatial omics technologies, successfully identifying biologically meaningful spatial multicellular patterns in axolotl brain regeneration, brain aging, liver zonation, Alzheimer’s disease, and colorectal cancer. Notably, FDPMining enables landmark-anchored pattern discovery around specific anatomical or pathological features such as blood vessels or amyloid plaques, among which applications to Alzheimer’s disease revealed previously inaccessible insights into the multicellular organization of these microenvironments. FDPMining offers a paradigm for quantitatively dissecting spatial heterogeneity in complex tissues, enabling more systematic mining, visualization, and interpretation of cellular organization across diverse biological conditions.

12 June 2025 in IEEE Transactions on Audio, Speech and Language Processing
Verified
Journal Article
Closed access
SWSheng Wang
Sheng Wang
EZEnwei Zhu
Enwei Zhu
FZFangyuan Zhao
Fangyuan Zhao
DBDechao Bu
Dechao Bu
JLJinpeng Li
Jinpeng Li
YZYi Zhao
Yi Zhao
Electronic Medical Records (EMRs), extensively recognized as a significant repository of clinical experience and medical knowledge, often jeopardize their utility by being typically penned in free text, resulting in under-structured information. This lack of structural organization poses a major hurdle in fully capitalizing on the medical data embedded in these EMRs. Medical information extraction (IE) can fill this gap by converting the clinic text to structured data. In this paper, we propose Marker LAttice Transformer (MAT), a strong framework for medical IE. This framework is composed of three separate models, each designed for a specific task: medical entity recognition (MER), medical relation extraction (MRE), and medical attribute extraction (MAE). All the models are deeply based on markers embedded in the input text, with which the models compute representations from bottom to top layers. This allows the representations to encode deep semantic information, leading to better outputs. In addition, we enhance the models by lattice-style incorporation of medical dictionary information, further pre-training on large-scale EMRs, and auxiliary inputs of medical departments and EMR sections. We evaluated MAT using the HwaMei-500 dataset, the most comprehensive and current collection of Chinese electronic medical records. The framework demonstrated exceptional performance, achieving $F_{1}$ scores of 93.0%, 71.5%, and 88.9% for MER, MRE, and MAE tasks respectively. These scores outperform the baseline by large margins and also surpass the results of recent state-of-the-art relation extraction models. As a medical IE framework, we provide a practical example for other information extraction efforts, particularly those focusing on Chinese electronic medical record data. In the spirit of collaborative research, we have publicly released our code and all relevant resources at https://github.com/WangSheng21s/MAT.