About

Weihsueh Chiu is associated with the Texas A&M College of Veterinary Medicine & Biomedical Sciences (VMBS), which is ranked as the No. 3 veterinary college in the United States and is recognized for its research in animal science and veterinary medicine. The college emphasizes a 'One Health' approach, recognizing the complex interactions between animal, human, and environmental health, and engages in cutting-edge basic, clinical, and translational research to address pressing scientific questions. While the specific research focus of Weihsueh Chiu is not detailed in the provided page text, the college's overall mission involves supporting collaborations that turn discoveries into proactive solutions for animal, human, and environmental health. The college also emphasizes innovative research, veterinary education, outreach, and emergency support services, reflecting a comprehensive commitment to advancing veterinary medicine and biomedical sciences.

Research topics

• Medicine
• Computer Science
• Biology
• Ecology
• Engineering
• Environmental science
• Environmental chemistry
• Environmental health
• Pathology
• Genetics
• Organic chemistry
• Geography
• Gerontology
• Chemistry
• Library science
• Biochemistry
• Veterinary medicine
• Management science
• Bioinformatics
• Family medicine
• Internal medicine
• Social psychology
• Pharmacology
• Mathematics

Selected publications

• Uncertainty-aware machine learning to predict non-cancer human toxicity for the global chemicals market

  Figshare · 2026-01-01

  datasetOpen access

  Predicted points of departure for reproductive/development and general non-cancer toxicity with 95% confidence intervals for 134,014 globally marketed chemicals predicted with "uncertainty-aware" machine learning models using random-forest based conformal prediction models.

  PublisherDOI

• Quantitative estimates of inter-individual variability for new approach methodologies-based systemic safety toolbox using a population-based human in vitro model

  Toxicological Sciences · 2026-03-23

  articleOpen access

  Next-generation risk assessment (NGRA) frameworks use new approach methodologies (NAMs) to support regulatory decisions without animal testing. Although NAM-based approaches are well established for hazard and dose-response assessment, inter-individual variability is still typically addressed using default uncertainty factors for inter-individual variability. This study evaluated an NAM-based strategy to quantify chemical-specific variability using a human cell model. We hypothesized that integrating chemical-specific variability data into NGRA would yield more protective risk estimates. Using 131 human lymphoblastoid cell lines (LCLs) from four European and African subpopulations, we assessed differences in cytotoxic responses to 53 substances, including industrial chemicals, pharmaceuticals, pesticides, and consumer-use compounds. Concentration-response testing (0.3 nM to 300 μM) data were analyzed using Bayesian modeling to calculate points of departure per cell line. Of the substances tested, 18 exhibited cytotoxic effects, enabling the derivation of chemical-specific variability factors. These factors were designated as toxicodynamic variability factors at the 5th percentile (TDVF05) because of the limited metabolic capacity of lymphoblast cell lines. The median TDVF05 was 3.8 (range 1 to 46), largely consistent with default assumptions. A genome-wide association study (GWAS) identified genomic loci, primarily containing transporter and metabolism genes, associated with variability in cytotoxicity, suggesting mechanistic bases for inter-individual differences. Overall, this study shows that human LCLs are a practical high-throughput in vitro model for quantifying inter-individual variability, strengthening confidence in NGRA risk predictions and supporting hypothesis generation on chemical-specific genetic and mechanistic drivers of human variability. However, cell-based systems have limited coverage of adverse effects and require careful alignment with in vivo dosimetry.

  PublisherOA PDFDOI

• Maternal PFAS transfer through lactation: dolphin milk reveals routes of early-life exposure

  Analytical and Bioanalytical Chemistry · 2026-02-03

  articleOpen access
  PublisherOA PDFDOI

• Application of a human bronchoepithelial—air–liquid interface model to assess respiratory hazard of VOCs using a benchmark concentration modeling approach

  Inhalation Toxicology · 2026-02-04

  article

  OBJECTIVE: human bronchial epithelial air-liquid interface (ALI) model. METHODS: A human bronchial epithelial cell line, 16HBE, was cultured at ALI and exposed to relevant concentrations of two representative VOCs, acrolein or formic acid, and matched filtered air (control) in a CelTox exposure system for two hours to replicate an acute inhalation exposure. Cells were allowed to recover for 24 h before cell lysate and culture media were collected for analysis. RESULTS: formic acid data produced BMCLs below existing regulatory exposure thresholds. CONCLUSION: tool to investigate VOC-induced effects on the airway and supports its utility in VOC safety evaluation.

  PublisherDOI

• Uncertainty-aware machine learning to predict non-cancer human toxicity for the global chemicals market

  Figshare · 2026-01-01

  datasetOpen access

  Predicted points of departure for reproductive/development and general non-cancer toxicity with 95% confidence intervals for 134,014 globally marketed chemicals predicted with "uncertainty-aware" machine learning models using random-forest based conformal prediction models.

  PublisherDOI

• Uncertainty-aware machine learning to predict non-cancer human toxicity for the global chemicals market

  Nature Communications · 2026-01-07 · 6 citations

  articleOpen access

  Humans are exposed to thousands of chemicals, yet limited toxicity data hinder effective management of their impacts on human health. High-performing machine learning models hold potential for addressing this gap, but their uncharacterized prediction performance across the wider range of chemicals undermines confidence in their results. We develop uncertainty-aware models to predict reproductive/developmental and general non-cancer human toxicity effect doses. Our well-calibrated models provide uncertainty estimates aligned with observed prediction errors and chemical familiarity. We predict toxicity with 95% confidence intervals for >100,000 globally marketed chemicals and identify toxicity and uncertainty hotspots. These results can be applied to inform decisions aimed at reducing potential human health impacts and guide targeted data generation and modeling efforts to reduce prediction uncertainty. Here, we show that enhancing transparency in prediction uncertainty provides key insights for building confidence in toxicity predictions, supporting the sound integration of machine learning-based predictions in chemical assessments.

  PublisherDOI

• How to organise a successful toxicology workshop? A participant perspective on the Collaboration to Harmonise the Assessment of Next Generation Evidence (CHANGE) workshop in Oslo, 18–20 June 2024

  Archives of Toxicology · 2025-05-15 · 1 citations

  articleOpen access

  To create a regulatory infrastructure for the effective use of NAMs, the CHANGE project aims to organise three workshops to Collaborate and Harmonise the Assessment of Next Generation Evidence. To better ensure the success of the CHANGE approach, project organisers invited a group of participants to provide feedback on the first workshop held in Oslo on 18-20 June 2024. This report represents the participants' perspective on the CHANGE working methodology and serves as a companion piece to the CHANGE organisers' publication "Improving how we use workshops when solving complex research problems: Reflections from the CHANGE Project", which provides a detailed description of the methodology, outputs, and conclusions from the workshop. The report includes feedback from most participants in response to the workshop evaluation as well as personal experiences from the authors. The workshop successfully facilitated stimulating engagement with a diversity of perspectives, though representation could be further broadened across sectors and geographies. Additionally, future workshops could refine the explanation of novel approaches to participants, as well as improve how information gathering is structured and formatted for feedback. Overall, participants were enthusiastic about CHANGE and believe the approach holds great promise in shaping future effective use of NAMs for chemical safety assessments. The report concludes with recommendations for follow-up workshops in 2025 and 2026, aiming to contribute to a regulatory infrastructure open to the acceptance and effective use of NAMs and to the use of similar workshops to address other emerging science policy issues.

  PublisherOA PDFDOI

• Exploring experiences of the regulatory toxicology system: system-level promoters and inhibitors of new approach methodologies

  Archives of Toxicology · 2025-09-09

  articleOpen access

  The transition from traditional animal-based approaches and assessments to New Approach Methodologies (NAMs) marks a scientific revolution in regulatory toxicology, with the potential of enhancing human and environmental protection. However, implementing the effective use of NAMs in regulatory toxicology has proven to be challenging, and so far, efforts to facilitate this change frequently focus on singular technical, psychological or economic inhibitors. This article takes a system-thinking approach to these challenges, a holistic framework for describing interactive relationships between the components of a system of interest. In this case, the regulatory toxicology system. We do so by analysing and interpreting a very large qualitative data set of experts' observations, collected in a 3-day interactive workshop and three follow-up online workshops with a heterogeneous sample of experts representing major actors from the global regulatory toxicology system. We identified leverage points (where a small change within a system can have a disproportionately large effect) in the six core aspects-infrastructure, processes, culture, technology, goals, and actors-in the regulatory toxicology system to facilitate the effective use of NAMs. Identified systematic leverage points include the need for a functioning incentive structure for effectively discovering, developing, validating and using NAMs within academia, regulation, and industry; and measures that prevent or mitigate unwanted effects of using NAMs that acknowledge clashes between scientific, regulatory, political and social processes. The results serve as a basis for follow-up activities that reflect on the actual effectiveness of these levers and that develop measures for the regulatory toxicology system.

  PublisherOA PDFDOI

• P30-56 Human vs. Cells vs. Machine: A comparative analysis between toxicological points of departure derived from quantitative read-across (qRAx), in vitro data, or in silico predictions

  Toxicology Letters · 2025-09-01

  article1st authorCorresponding
  PublisherDOI

• Derivation of Human Toxicokinetic Parameters and Chemical-Specific Adjustment Factor of Citrinin Through a Human Intervention Trial and Hierarchical Bayesian Population Modeling

  Toxins · 2025-07-31

  articleOpen accessSenior author

  Background: Citrinin (CIT) is a mycotoxin produced by various fungi contaminating stored cereals and fruits. While biomonitoring and food occurrence data indicate widespread exposure, its public health risks remain unclear due to the lack of human toxicokinetic (TK) data. Methods: A UHPLC-MS/MS method was validated for CIT quantification in capillary blood (VAMS Mitra® tips), feces, and urine obtaining LLOQs ≤ 0.05 ng/mL. A human TK study was conducted following a single oral bolus of 200 ng/kg bw CIT. Individual capillary blood (VAMS Mitra® tips), feces, and urine samples were collected for 48 h after exposure. Samples were analyzed to determine CIT’s TK profile. Results: TK modeling was performed using a multi-compartmental structure with a hierarchical Bayesian population approach, allowing robust parameter estimation despite the lack of standards for CIT metabolites. Conclusions: The derived TK parameters align with preliminary human data and significantly advance CIT exposure assessment via biomonitoring. A human inter-individual toxicokinetic variability (HKAF) of 1.92 was calculated based on the derived AUC, indicating that EFSA’s current default uncertainty factor for TK variability is adequately protective for at least 95% of the population.

  PublisherOA PDFDOI

Recent grants

• Project 4

  NIH · $14.7M · 2022–2027

• IMSD at Texas A&M University: Initiative for Maximizing Student Diversity in Biomedical Sciences

  NIH · $1.2M · 2020–2025

• Texas A&M Center for Environmental Health Research (TiCER)

  NIH · $20.5M · 2019–2029

Frequent coauthors

• Ivan Rusyn

  Texas A&M University

  118 shared

• Kathryn Z. Guyton

  National Academies of Sciences, Engineering, and Medicine

  46 shared

• Fred A. Wright

  42 shared

• Yu‐Syuan Luo

  National Taiwan University

  25 shared

• Thomas J. McDonald

  Texas A&M Health Science Center

  25 shared

• Cheryl Siegel Scott

  National Center for Environmental Assessment (EPA)

  23 shared

• Joseph A. Cichocki

  Vertex Pharmaceuticals (United States)

  23 shared

• Suji Jang

  Texas A&M University

  19 shared