About

Dr. Fusheng Wang is an assistant professor at the Department of Biomedical Informatics and the Department of Computer Science at Stony Brook University. He received his Ph.D. in Computer Science from the University of California, Los Angeles in 2004, and his M.S. and B.S. in Engineering Physics from Tsinghua University, China, in 1997 and 1994 respectively. Prior to joining Stony Brook University, he was an assistant professor at Emory University and a research scientist at Siemens Corporate Research from 2004 to 2009. His research interests include Big Data Management and Analytics, Spatial and Temporal Data Management and Analytics, Medical Imaging Informatics, Heterogeneous Data Management and Integration, Clinical Natural Language Processing, and Data Semantics and Standardization. His work is sponsored by NSF, NIH, Amazon, and Google. Dr. Wang has been recognized with the NSF CAREER Award in 2014.

Research topics

• Artificial Intelligence
• Computer Science
• World Wide Web
• Computational biology
• Speech recognition
• Human–computer interaction
• Computer vision
• Cell biology
• Anatomy
• Biology

Selected publications

• Physics-Informed Complex Natural Resonance Frequency Correction

  IEEE Antennas and Wireless Propagation Letters · 2025-06-10

  article

  Complex natural resonance frequency is only determined by target's intrinsic properties, and independent of incident angle and target's attitude, making target recognition based on complex natural resonance frequency a promising solution. However, the existing extraction methods inevitably introduce calculation errors, resulting in inaccurate extraction of it, which limits the improvement of recognition accuracy. To address these issues, Physics-Informed Complex Natural Resonance Frequency Correction (PI-CNRFC) is proposed. Firstly, a loss function based on target's resonance scattering mechanism is designed. Physics is used to drive network to extract features, strengthening the physical constraints on the corrected data. Secondly, by introducing the idea of Complex-valued Neural Network (CVNN), a Complex domain- Variational AutoEncoder (C-VAE) is designed, which considers the correlation between the real and imaginary parts of complex data,thereby excavating more internal features of data. Experimental results verify the effectiveness of the proposed method. Compared with the original data, the relative errors are reduced by 18.02% for the real part and 0.49% for the imaginary part. improving the consistency between corrected data and target's resonance scattering mechanism.

  PublisherDOI

• THU-255 The safety and efficacy of PD-1 antibody combined with pegylated interferon-α therapy to promote the clinical cure in nucleoside (acid) analogues-suppressed chronic hepatitis B patients

  Journal of Hepatology · 2025-05-01

  article
  PublisherDOI

• Efficient and Accurate Spatial Queries Using Lossy Compressed 3D Geometry Data

  IEEE Transactions on Knowledge and Data Engineering · 2025-02-26 · 1 citations

  articleSenior author

  3D spatial data management is increasingly vital across various application scenarios, such as GIS, digital twins, human atlases, and tissue imaging. However, the inherent complexity of 3D spatial data, primarily represented by 3D geometries in real-world applications, hinders the efficient evaluation of spatial relationships through resource-intensive geometric computations. Geometric simplification algorithms have been developed to reduce the complexity of 3D representations, albeit at the cost of querying accuracy. Previous work has aimed to address precision loss by leveraging the spatial relationship between the simplified and original 3D object representations. However, this approach relied on specialized geometric simplification algorithms tailored to regions with specific criteria. In this paper, we introduce a novel approach to achieve highly efficient and accurate 3D spatial queries, incorporating geometric computation and simplification. We present a generalized progressive refinement methodology applicable to general geometric simplification algorithms, involving accurate querying of 3D geometry data using low-resolution representations and simplification extents quantified using Hausdorff distances at the facet level. Additionally, we propose techniques for calculating and storing Hausdorff distances efficiently. Extensive experimental evaluations validate the effectiveness of the proposed method which outperforms state-of-the-art systems by a factor of 4 while minimizing computational and storage overhead.

  PublisherDOI

• Towards Inpatient Discharge Summary Automation via Large Language Models: A Multidimensional Evaluation with a HIPAA-Compliant Instance of GPT-4o and Clinical Expert Assessment

  medRxiv · 2025-04-04 · 1 citations

  preprintOpen access

  Abstract Large language models (LLMs) have demonstrated potential to automate clinical documentation tasks that may reduce clinician burden, such as generation of hospital discharge summaries. Prior research used older LLMs and limited data, raising concerns about fabrications and omissions. In this study, we evaluated the automatic generation of inpatient Internal Medicine discharge summaries using a HIPAA-compliant Microsoft Azure instance of OpenAI’s GPT-4o. Both human-written and AI-generated discharge summaries were scored by Internal Medicine hospital faculty for quality, readability/conciseness, factuality and completeness, presence of hallucinations/omissions and their impact on safety, and compared with the actual discharge summaries. Our results showed that the AI-generated discharge summaries significantly outperformed actual human written summaries in both quality and readability/conciseness and were comparable to humans in factuality and completeness, with a minimal cost.

  PublisherOA PDFDOI

• Cell type populations for 3D anatomical structures of the Human Reference Atlas

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-08-20 · 2 citations

  preprintOpen access

  The human body contains ~27-36 trillion cells of up to 10,000 cell types (CTs) within a volume of ~62-120 liters (males) and 52-89 liters (females). The Human Reference Atlas (HRA) v2.3 provides a quantitative 3D framework of CTs across 73 reference organs and 1,283 3D anatomical structures (ASs). The HRA Cell Type Population (HRApop) effort quantifies CTs per AS using high-quality single-cell (sc) data processed through scalable, reproducible workflows and cell type annotation (CTann) tools. HRApop v1.0 includes reference CT populations for 73 ASs (112 when sex-specific) using 662 datasets spatially registered to 230 locations across 17 organs (31 when sex-specific). For 558 sc-transcriptomics datasets (11,042,750 cells), CTs and biomarker expression were computed using Azimuth, CellTypist, and popV. To test generalizability, 104 sc-proteomics datasets (16,576,863 cells) were integrated. In total, HRApop includes 27,619,613 cells. HRApop can be used to predict (1) CT populations for 3D volumes in the human body and (2) the spatial origin of a tissue block, given a CT population. Data and code are at cns-iu.github.io/hra-cell-type-populations-supporting-information.

  PublisherOA PDFDOI

• Multi-omics dissection of metabolic dysregulation associated with immune recovery in people living with HIV-1

  Journal of Translational Medicine · 2025-01-31 · 8 citations

  articleOpen access

  BACKGROUND: Despite the success of antiretroviral therapy (ART) in suppressing HIV-1 replication, some people living with HIV-1 (PLWH) fail to achieve an optimal recovery of CD4 T cells, and precise metabolic regulation underlying immune recovery remained poorly understood. METHODS: In this cross-sectional study, mass spectrometry was used for quantitative analysis of plasma metabolome and lipidome in 24 treatment-naïve PLWH (TNs), 33 immunological responders (IRs), 35 immunological non-responders (INRs), and 16 healthy controls (HCs). The data were analyzed using the Mann-Whitney U-test, Kruskal-Wallis test, Spearman correlation, and LASSO regression analysis. RESULTS: Significant metabolic dysregulation was observed in TNs, IRs and INRs compared to HCs. In TNs, metabolomic analysis revealed increased levels of 3-hydroxyoctanoic acid, 3-oxododecanoic acid, 5-hydroxy-L-tryptophan, 5-hydroxyindoleacetic acid, L-kynurenine, oleoylcarnitine, and pseudouridine that were positively correlated with CD8 T cell activation and inflammation-related markers, and decreased levels of phosphorylcholine, ribothymidine, and thymine that were negatively correlated. Notably, 3-hydroxyoctanoic acid and thymine were consistently associated with CD4 T cell counts and inflammation-related markers in PLWH, regardless of ART. Pathway analysis uncovered the biosynthesis of unsaturated fatty acids as the major dysregulated pathway in TNs, IRs, and INRs, while primary bile acid biosynthesis was the dysregulated pathway specifically in INRs. Lipidomic analysis indicated higher plasma triacylglycerols, free fatty acids, ceramide, and monosialodihexosyl gangliosides (GM3) in TNs, IRs, and INRs compared to HCs. Pathway enrichment and differential correlation analyses underscore perturbed systemic lipid metabolism in treatment response to ART, possibly mediated by host-commensal metabolic interactions. Ultimately, we identified two panels, one consisting of 9 metabolites and another of 8 lipids, that can effectively distinguish INRs from IRs. CONCLUSIONS: Metabolic aberrations induced by chronic HIV-1 infection persist and do not recover with ART. Abnormal primary bile acid biosynthesis pathway and levels of DHA-containing lipids are closely associated with CD4 T cell recovery. These finding provide new intervention targets to achieve better immune recovery in PLWH.

  PublisherOA PDFDOI

• Demographic disparities, temporal trends, and geographic patterns of HPV vaccination on Long Island, New York: A comprehensive analysis of immunization registry data (2012–2023)

  Human Vaccines & Immunotherapeutics · 2025-04-15 · 1 citations

  articleOpen accessSenior authorCorresponding

  birthday and 24.95% of them received their second dose timely, given 6 to 12 months after the first dose. Spatially, eastern LI consistently holds a higher HPV vaccination rate than northwestern LI due to regional disparities. The rising trend in HPV vaccination coverage on LI with a modest drop around 2020 indicates the likely impact of COVID-19 pandemic. This study suggests the need for increased focus on regions with lower vaccine uptake rates for preventing HPV-related cancers.

  PublisherDOI

• Carbon ion beam-induced radiation hormesis in Bupleurum chinense DC.: Insights from growth, physiological, and metabolomic analyses for increased bioactive substances

  Plant Stress · 2025-09-23

  articleOpen access

  • The yield of B. chinense remained stable after CIB pretreatment, even though seedling growth was temporarily suppressed. • The antioxidant system and photosynthetic system play crucial roles in the ROS-mediated stress priming and radiation hormesis of B. chinense. • Pretreatment with 50 Gy of CIB significantly increased total flavonoids, saikosaponins a and c, and several pharmacologically active substances in B. chinense. • Metabolomics revealed the reprogramming of flavonoid and terpenoid biosynthesis in B. chinense induced by CIB radiation. • CIB pretreatment represents a promising and environmentally friendly approach to enhance secondary metabolites in medicinal plants. Bupleurum chinense DC ( B. chinense ) is an important medicinal plant widely used in Asian countries. However, the efficacy of its medicinal components has medicinal components during the domestication process from wild to cultivated lines. Studies have shown that appropriate stress, such as ionizing radiation, can promote the accumulation of metabolites in medicinal plants. Nevertheless, the effects of ionizing radiation on B. chinense remains unclear. In this study, we systematically investigated the stimulatory effect of carbon ion beams (CIB) pretreatment on the growth, physiology, and accumulation of secondary metabolites in B. chinense . Although CIB irradiation inhibited the seed germination and survival rates, by the age of 4 months, the plant height and leaf area of the irradiated group had recovered to the levels comparable to the control. The enhanced growth performance during later developmental stages may be attributed to radiation hormesis and ROS-mediated regulation in antioxidant system and the photosynthetic system. One and two years after irradiation at doses of 50 and 100 Gy, no significant differences in root biomass were observed compared to the control group. Metabolically, the content of total flavonoid, saikosaponin a and saikosaponin c were significantly increased following 50 Gy irradiation. Further metabolomic analysis revealed that intermediate metabolites in the flavonoid and terpenoid biosynthetic pathways were significantly up-regulated in the 50 Gy irradiation group. Compounds with pharmacological activity also accumulated in large quantities after irradiation. These results suggest that pretreatment with 50 Gy CIB irradiation could serve as a potential method to promote the accumulation of secondary metabolites in B. chinense . This finding provides strong support for the application of physical radiation technology to enhance the production of secondary metabolites in medicinal plants, offering new avenues for the cultivation of high-quality Chinese herbal medicines.

  PublisherDOI

• MicroRNA-126-3p as a predictive biomarker for patients with primary biliary cholangitis refractory to ursodeoxycholic acid

  World Journal of Gastroenterology · 2025-08-18 · 1 citations

  articleOpen access

  BACKGROUND Ursodeoxycholic acid (UDCA) is the first-line therapeutic agent for primary biliary cholangitis (PBC). However, a subset of patients exhibit a suboptimal response to UDCA, and reliable predictive biomarkers remain elusive. Studies have implicated plasma microRNAs (miRNAs) in the pathophysiological progression of PBC, with certain miRNAs demonstrating potential as diagnostic and disease progression biomarkers. However, biomarkers capable of predicting the therapeutic efficacy of UDCA have not yet been identified. AIM To investigate differentially expressed miRNAs in PBC patients with divergent UDCA treatment responses and to explore potential biomarkers that predict treatment response in PBC. METHODS Plasma samples from treatment-naive PBC patients receiving ≥ 1 year of standard UDCA treatment were collected. Efficacy was evaluated using the Paris I criteria. Patient samples were divided into discovery group (n = 10) and validation group (n = 30), with further stratification of patients into drug-resistant and drug-sensitive (DS) cohorts. Next-generation sequencing and quantitative real-time polymerase chain reaction were used to screen, functionally analyze, and validate the pre-treatment miRNA profiles of the treatment groups. RESULTS Forty-nine miRNAs were differentially expressed between the two groups before UDCA treatment (N = 40). MiR-22-5p and miR-126-3p were highly expressed in the DS group before treatment (P < 0.001), whereas miR-7706 exhibited a low expression (P = 0.017). Post-treatment, miR-126-3p maintained low expression in the drug-resistant group (P = 0.003), but showed elevated levels in the DS group (P < 0.001). Logistic regression analysis identified miR-126-3p expression (odds ratio = 34.32, 95% confidence interval: 1.95-605.40, P = 0.016) as a significant factor influencing UDCA treatment response, while miR-22-5p (P = 0.990) and miR-7706 (P = 0.157) showed no significant association. MiR-126-3p levels were negatively correlated with total bilirubin (r = -0.356, P = 0.005) and immunoglobulin G levels (r = -0.311, P = 0.015). The area under the receiver operating characteristic curve was 0.891 (P = 0.0003, 95% confidence interval: 0.772-1.000) with a sensitivity of 82.4% and a specificity of 84.6%. CONCLUSION Plasma miRNA expression profiles are heterogenous in patients with PBC with differential responses to UDCA therapy. MiR-126-3p demonstrates predictive potential for a suboptimal response to UDCA in patients with PBC.

  PublisherDOI

• Pyrite-enhanced coal spontaneous combustion: Insights from experiments and molecular simulations

  Fuel · 2025-02-20 · 13 citations

  article
  PublisherDOI

Recent grants

• CAREER: High Performance Spatial Queries and Analytics for Spatial Big Data

  NSF · $414k · 2015–2021

• CAREER: High Performance Spatial Queries and Analytics for Spatial Big Data

  NSF · $87k · 2014–2015

• Computational pathology software for integrative cancer research with three-dimensional digital slides

  NIH · $1.1M · 2019–2023

Frequent coauthors

• Jun Kong

  North Dakota State University

  94 shared

• Joel Saltz

  63 shared

• George Teodoro

  Universidade Federal de Minas Gerais

  41 shared

• Sina Rashidian

  Stony Brook University

  41 shared

• Mary Saltz

  37 shared

• Richard N. Rosenthal

  Stony Brook University

  33 shared

• Kayley Abell-Hart

  Stony Brook University

  32 shared

• Zheng Zhang

  Shenzhen Third People’s Hospital

  31 shared

Education

• Ph.D., Computer Science

  University of California Los Angeles

  2004

Awards & honors

• NSF CAREER Award (2014)