ICCVAM Biennial Report 2024-2025

The EPA Ecotoxicology Knowledgebase (ECOTOX) provides public access to comprehensive information on adverse effects of single chemical stressors to ecologically relevant aquatic and terrestrial species. ECOTOX includes curated data from over 55,000 references in the scientific literature and over 1.2 million test records, covering more than 14,000 aquatic and terrestrial species and 13,000 chemicals.

There were eight updates to ECOTOX during 2024–2025. Recent additions include information on 6PPD-quinone, 6PPD-alternatives, arsenic, cyanotoxins, metolachlor, quizalofop, and per- and polyfluoroalkyl substances (PFAS). In addition to the data updates, the Explore Publications module launched in 2024 enables queries by citation information (author, title, ECOTOX ECOREF number, DOI) and displays toxicity data by reference. In 2024 and 2025, ECOTOX had an average of over 16,000 monthly page views and over 2,000 active monthly users. An ECOTOX training session was presented in April 2024 as part of a broader in-person EPA NAMs tools training workshop. Additional training resources for ECOTOX are available on the EPA NAMs Training webpage.

The NIH Integrated Chemical Environment (NICE) provides data and tools to help develop, assess, and interpret chemical safety tests. The March 2024 NICE v. 4.0.2 release improved support for chemical name searching in the NICE REST API. The September 2024 NICE v. 4.1 release added versioning information to datasets and added applicability domain information for chemical property predictions from the Open (Quantitative) Structure–activity/property Relationship App (OPERA). The July 2025 NICE v. 4.2 release updated curated high-throughput screening (cHTS) data to align with data in the EPA ToxCast invitrodb v. 4.2 release. Other updates and improvements to NICE datasets during 2024–2025 include:

• Links to the NTP Chemical Effects in Biological Systems database and the EPA CompTox Chemicals database in data downloads
• Data annotation aligned with Open Biological and Biomedical Ontology Foundry ontologies
• Ability of users to download ClassyFire (Djoumbou Feunang et al. 2016) chemical taxonomies for over one million chemicals
• A new “PFAS” Chemical Quick List; a “Demo List of Chemicals” Chemical Quick List was also added to help new users explore NICE functionality
• An updated “RoC Classifications” Chemical Quick List that includes new data from the NTP’s 15th Report on Carcinogens
• An update of the “Tox21” Chemical Quick List
• Continued harmonization of skin sensitization and eye irritation data

ICCVAM has identified a need to enhance access to information about NAMs that have been qualified and validated for regulatory application. The Collection of Alternative Methods for Regulatory Application (CAMERA) was developed in response to that need.

CAMERA is an interactive database and user interface that provides online access to validated alternative methods for U.S. regulatory use and other contexts of use. It is being developed by NICEATM under guidance from an ICCVAM steering committee. CAMERA is a central hub and unified resource of validated alternative methods that enhances access to validation study reports, data, protocols, standard operating procedures, and information on regulatory guidance. It provides access to test method details, such as chemical applicability domain, related key events, and reference chemicals. Users can filter searches by alternative method type, defined approach, test method endpoint, and regulatory guidance.

As of the end of 2025, CAMERA was in development, with some pages still having limited functionality. The database includes 27 NAMs that cover endpoints including skin irritation and corrosion, eye irritation, and skin sensitization. Updates planned for early 2026 will add tooltips to improve user guidance on the website, revise the chemicals and performance metrics tables, and add new definitions to the glossary.

BioBricks (Insilica) is an application that makes large bioinformatics databases programmatically accessible. Because of the potential usefulness of this tool for toxicological data acquisition and analysis, NICEATM began working with Insilica in 2022 to provide support and testing of the BioBricks platform. The NICEATM team contributed to building numerous bricks (units that encapsulate datasets), including those derived from NICE datasets. NICEATM also supported the testing and expansion of the BioBricks interface through activities such as the development of a Python version. A complete description of BioBricks has been published in Frontiers in Artificial Intelligence (Gao et al. 2025), with NICEATM team members serving as coauthors. All databases integrated into BioBricks are fully and freely available; the code is open-source and available via a GitHub repository.

In 2025, NTP published a groundbreaking online resource called the Rodent Ultrastructure Atlas. This digital tool provides detailed electron microscope (ultrastructural) images of subcellular structures in rodent tissues, offering researchers an essential reference for identifying changes caused by diseases or chemical exposures. Understanding "normal" cellular and tissue ultrastructure is critical for diagnosing abnormalities, and the Atlas serves as a baseline for comparison.

While light microscopy is commonly used for tissue evaluation, it cannot reveal structures within cells. In contrast, electron microscopy reveals subcellular structures, allowing researchers to pinpoint how a disease or chemical affects specific parts of cells. By providing reference photos, the Atlas can help scientists determine whether a compound has reached its cellular target or disrupted tissue and organ function.

This first-of-its-kind resource is freely accessible to researchers worldwide through the NTP website. Users can explore content by organ systems, such as endocrine, immune, or respiratory, and zoom in for detailed views of ultrastructural components. Each chapter combines overview images with detailed, high-resolution, ultrastructure images, making it easy for scientists to study tissues and identify cell components.

By providing researchers with a comprehensive library of high‑quality images of healthy tissues, the Atlas reduces the need to generate new comparison samples from additional animals. Researchers can evaluate whether observed changes are meaningful by referencing existing images, helping minimize reliance on new in vivo studies. Although this Atlas currently focuses on normal ultrastructure, it will continuously expand to include examples of abnormal subcellular changes linked to toxicity and diseases such as cancer. Through collaboration with toxicologic pathologists globally, the NTP team will continue to expand and improve this resource, advancing research in toxicology and pathology. 

Increases in the incidence of neurodevelopmental disorders, such as attention deficit/hyperactivity disorder, dyslexia, and autism spectrum disorder, have prompted scientific interest in the potential contribution of environmental toxicants to the development of these disorders. To support these investigations, the NIEHS Division of Translational Toxicology launched the Developmental NeuroToxicity Data Integration and Visualization Enabling Resource (DNT-DIVER). DNT-DIVER allows users to analyze, compare, and visualize multiple DNT assays in an interactive web-based application. The application provides public access to and visualization of all data from compounds screened by the division’s Developmental Neurotoxicity Health Effects Innovation Program. The current version (v. 2.2) includes different tabs for experimental design summary, quality control, chemical-specific concentration–response curves, ranking of chemical toxicity per laboratory/assay, and comparison of results across assays. DNT-DIVER will also accept data from novel screening assays as they become available and are accepted by the international community of researchers investigating DNT.

In recent decades, the prevalence rates of neurodevelopmental disorders have increased. While improved diagnosis plays a role, environmental factors, such as chemical exposure during pregnancy, may also contribute. Despite this concern, data regarding DNT potential of most chemicals remain limited, and DNT testing is not typically required during chemical registration or reregistration. Current DNT testing methods rely on animal studies such as those described in Test Guideline 426 issued by the Organisation for Economic Co-operation and Development (OECD). Such studies are expensive, time-consuming, and ethically challenging, and they provide limited insight into how these substances affect human brain development.

To address these challenges, international researchers have developed in vitro NAMs to evaluate DNT more efficiently. This effort led to the creation of the DNT in vitro battery, a set of lab-based tests designed to assess key brain developmental processes without using live animals. OECD has published initial recommendations on how to use DNT in vitro battery data; however, for these methods to become officially recognized as OECD guidance, they must undergo testing in additional laboratories worldwide.

The FDA Human Foods Program has begun in-house validation of some DNT in vitro battery assays developed by organizations such as EPA, in order that they can be run within the Human Foods Program with the same reliable results. However, this transfer process faces challenges. For example, many required cell models are not commercially available. Additionally, there is no agreed list of "performance compounds" to reliably evaluate performance across laboratories. Successfully addressing these challenges will improve DNT testing worldwide, promoting consistency, reducing animal use, and advancing understanding of how chemicals may affect brain development, a critical step in protecting public health, especially for children.

EPA and NICEATM researchers (Karmaus et al. 2026) reviewed the consistency of repeated traditional animal toxicity studies used in safety assessments. They examined published analyses covering various safety tests including eye and skin irritation, acute systemic toxicity, cancer studies, and developmental toxicity. The results showed that replicability, getting the same result when repeating a study, varied widely depending on the type of test and how the results are interpreted. For instance, a study could report a dose that caused toxicity, or it could be used in a categorical classification format with varying levels of concern. One recurring finding was that extreme results (finding that the chemical was either highly toxic or nontoxic) were more consistent than moderate effects. Evaluating replicability was difficult because studies are not frequently repeated for the same chemical, and the approach used in a subsequent study is not always the same as the approach used in the first. This variability comes from multiple sources, including differences in animal lines, experimental procedures, natural biological variation, and analysis methods. These findings are important because regulatory agencies are increasingly accepting new test methods that do not use animals, but these alternatives are often required to perform "as well as or better than" traditional animal test methods. Understanding that animal studies themselves have significant variability helps set realistic performance standards for new methods and suggests that some alternative approaches may be more reliable than the animal tests they are meant to replace.

Reviewing in vivo toxicity studies to assess the risk chemical exposures pose to human health is an important process. A foundational part of this process involves extracting and organizing information into a structured format. However, the scientific language used across studies may vary widely, requiring substantial manual work to standardize descriptions before the data can be analyzed effectively. To address this, EPA and NIEHS scientists developed and applied a semiautomated workflow to map pre-existing controlled vocabularies (standardized systems of terminology) to ensure consistency in how toxicological effects are described.

In this study, the team reviewed over 40,000 toxicological data points extracted from prenatal developmental toxicity studies conducted by NTP and the European Chemicals Agency. They used a harmonized controlled vocabulary system that combined terms from three sources: Unified Medical Language System codes as implemented in the EPA Toxicity Reference Database; German Federal Institute for Risk Assessment DevTox harmonized terms; and endpoint vocabularies used by OECD. Leveraging a programmatic approach, standardized terms were successfully matched to 75% of the NTP data and 57% of the European Chemicals Agency data. Approximately half of the standardized matches required manual review to ensure accuracy. Data points that could not be matched automatically were often too general or required human judgment to interpret.

This semiautomated workflow saved over 350 hours of manual work and created useful resources, including reproducible code for implementing the mapping workflow and a comprehensive, accessible dataset of harmonized vocabulary for developmental and reproductive toxicity. This crosswalk effort can be readily adapted to other study types. The results demonstrate how automation can advance the organization of toxicological data by facilitating semantic interoperability, an essential prerequisite for integrating legacy datasets to support analyses and decision-making.

The androgen receptor system is a critical component of the endocrine system that regulates male characteristics and reproductive health. Endocrine-disrupting chemicals can interfere with this receptor during critical developmental periods, potentially causing reproductive disorders, impaired fertility, and altered sexual development. The FDA National Center for Toxicological Research has developed the Molecules with Androgenic Activity Resource, the first comprehensive open-access database for assessing chemicals that may disrupt the androgen receptor (Dong et al. 2024).

The research team collected and carefully organized 125,519 data points on androgenic activity for 13,648 different chemicals from multiple public databases and scientific literature. These data were produced by 923 different laboratory tests, including binding assays, cell-based tests, and animal studies.

The Molecules with Androgenic Activity Resource addresses a major challenge in chemical safety assessment; while extensive test data exist, the data are scattered across different sources, reported in various formats, and difficult to access and compare. This resource addresses this problem by providing a centralized, user-friendly web platform where researchers, regulators, and industry professionals can easily search for chemicals and find comprehensive information about their potential to disrupt androgen receptors. The database includes detailed information about each chemical's structure, test results, and the methods used to generate the data. By making this information freely available and easy to use, the Molecules with Androgenic Activity Resource enables better risk assessment of chemicals and supports the development of predictive models to identify potentially harmful substances before expensive animal testing is required.

Animal tests have traditionally been used to identify acute inhalation toxicants and define safety standards that protect human health. NAMs that include in vitro and computational approaches have been proposed as complementary resources. These approaches have the potential to identify toxicants and elucidate toxicity mechanisms in a manner that is more relevant to human biology than animal models. However, developing and evaluating these NAMs requires robust, well-curated, and chemically diverse reference data.

Using data from open-access sources, NICEATM compiled a database of in vivo rat acute inhalation data for approximately 1,200 unique substances. In addition to LC50 values (exposure concentration of a toxic substance estimated to be lethal to half of the test animals), metadata collected for each entry included exposure type, exposure route, species, sex, and number of animals tested, when available. The chemical set was found to represent a diversity of physicochemical properties and functional use categories represented in the EPA Chemical and Products Database. EPA hazard categories (e.g., nontoxic, toxic, highly toxic) were assigned based on LC50 values. While all hazard categories are represented in the database, about two-thirds of the chemicals were classified as Category IV (least toxic/nontoxic). The final curated database contains 2,565 entries for 1,209 unique substances. Of these, 1,025 unique chemicals (2,109 entries) have chemical structures standardized for quantitative structure–activity relationship (QSAR) modeling.

A variability analysis was conducted on the 231 chemicals with two or more LC50 values by calculating mean absolute deviation, a measure of variability based on the absolute distance between each data value and the mean of the dataset. Principal component analysis was then used to evaluate any associations with inhalation LC50 variability, but no associations were found between LC50 values and either physicochemical properties or functional use categories. Conditional probabilities were also calculated for these chemicals to determine the probability that a chemical would be assigned a specific EPA hazard category given that it was previously assigned the same or another category on the basis of an LC50 from another study. This analysis showed Category IV categorization to be most consistent and Category III (mildly toxic) to be least consistent. The database can be downloaded from NICE, and a paper describing the database is in preparation.

PFAS are manufactured chemicals known for their widespread use and environmental concerns, such as persistence, bioaccumulation, and toxicity. While some PFAS have been thoroughly studied, most lack detailed hazard profiles. To address this, EPA researchers developed a systematic categorization process to organize PFAS based on their chemical structures, enabling better prioritization for future research and risk assessment (Patlewicz et al. 2025).

Using the official PFAS definition outlined under the Toxic Substances Control Act, the study initially identified 13,054 PFAS from the EPA Distributed Structure-Searchable Toxicity database. For the 563 PFAS listed in the nonconfidential Toxic Substances Control Act inventory, plausible chemical degradation products were simulated, adding 2,484 degradants to the list. Altogether, the researchers compiled a comprehensive dataset of 15,538 PFAS.

Each PFAS was assigned to primary categories based on OECD structure-based classifications and further divided into secondary categories by chain length (shorter or longer than seven carbons). These secondary groups were refined using chemical fingerprints, ensuring a balance between category size and structural similarity, resulting in 128 terminal structural categories. These categories were used to identify representative chemicals for data collection based on toxicity data gaps, environmental monitoring presence, and links to manufacturers/importers.

This innovative approach provides an organized framework for selecting PFAS for further toxicity testing. It has the potential to support hazard assessment, read-across methodologies, and the development of predictive tools such as QSAR models. These efforts will help address the lack of data for the majority of PFAS and improve understanding of their environmental and human health risks.

Drug-induced liver injury (DILI) is a serious safety concern that can lead to drug failure and patient harm. The FDA National Center for Toxicological Research's DILIrank database supports the development of NAMs to predict whether an investigational drug might cause DILI. The original database from 2009 contained information on approximately 1,000 drugs; DILIrank v. 2.0 (Olubamiwa et al. 2024) expands the database to 1,336 drugs approved through 2021, representing a 29% increase.

DILIrank categorizes drugs into four groups based on their liver injury risk: Most-DILI-Concern, Less-DILI-Concern, No-DILI-Concern, and Ambiguous-DILI-Concern. As part of the database update, researchers reviewed recent medical case reports and FDA drug labels, finding that some drugs previously considered safe now have associated documented liver injury cases. Interestingly, the analysis revealed that after the FDA issued specific liver safety guidance in 2009, newly approved drugs showed significantly improved safety profiles, with 39% classified as No-DILI-Concern compared to lower percentages in earlier periods. Cancer drugs (antineoplastic agents) continue to show the highest liver injury risk. The database also showed that drugs approved through expedited FDA programs in place between 1992 and 2008 had higher liver injury rates; however, many of these drugs were for treatment of life-threatening conditions, so higher liver injury rates were considered acceptable. This comprehensive resource will help researchers develop better models to predict liver toxicity and support safer drug development in the future.

Drug-induced kidney injury affects more than 1.5 million people in the U.S. each year, but current methods for detecting kidney damage often fail to identify problems until significant kidney function has already been lost. To help improve drug safety testing, FDA National Center for Toxicological Research researchers (Conor et al. 2024) created the Drug-Induced Renal Injury List, a comprehensive list of drugs with associated kidney damage risk levels. The research team compiled information from the scientific literature and FDA drug labels to create the list, which includes 317 orally administered drugs spanning all major drug categories. Of these drugs, 171 have potential to cause kidney injury, while 146 do not. This carefully curated list provides researchers with a valuable reference tool for developing NAMs that can potentially detect kidney toxicity earlier in the drug development process without relying solely on traditional animal models. Specifically, this list can be used to train and validate new predictive models, including computer-based approaches and advanced cell-based systems.

Key characteristics of carcinogens (KCCs) provide a foundation for organizing data on cancer mechanisms, with the potential to better characterize cancer hazard. To apply this framework to HTS assay data, a workgroup mapped assay data from EPA ToxCast invitrodb v. 4.2 to KCCs (Cuomo et al. 2026). The workgroup was made up of experts with diverse subject matter knowledge from NIEHS; EPA; California EPA; University of California, Berkeley; Texas A&M University; and the International Agency for Research on Cancer. Previous KCC assay mappings using the International Agency for Research on Cancer and NIH Integrated Chemical Environment databases served as the starting point for this work. Reviewers developed a harmonized mapping approach to relate different assay technologies and endpoints to carcinogenicity. The workgroup’s approach was to (1) clarify the contexts of the defined KCCs, considering directionality and biological coverage of the in vitro HTS assay space; (2) develop a tiered approach to mapping assays to integrate key considerations about the direct relevance of assay platforms to carcinogenesis; and (3) map relevant assays to one or more KCC, each with a detailed documented rationale for the attribution. Of the 1,570 assay endpoints available in the database, 838 were identified as mappable to at least one KCC. For each mapped KCC, assays were assigned as either “Tier A”, assays measuring a direct effect on the biology defined by a KCC or “Tier B”, assays measuring an indirect or downstream effect on the biological process underlying a KCC. This comprehensive mapping, validated through thorough manual review and documentation by a panel of experts, enhances the utility of HTS data for integration into carcinogenicity evaluation and ultimately contributes to a deeper understanding of the complex mechanisms underlying chemical-induced cancer. This effort will be incorporated into NICE and the International Agency for Research on Cancer Monographs for use in weight-of-evidence analyses and other integrative approaches for the identification of cancer hazards.

The Cancer Complexity Knowledge Portal is a free, open-access platform funded by the NCI’s Division of Cancer Biology and maintained by Sage Bionetworks through the Multi-Consortia Coordinating Center. NCI developed and is responsible for keeping the resource current. The portal serves as a central discovery hub for data, computational tools, publications, grants, and researcher profiles generated by several NCI-funded interdisciplinary research consortia. The portal is designed for a broad audience including cancer researchers, trainees, patient advocates, and funders. Visitors can search across all resource types simultaneously or browse by consortium, research theme, or cancer type. Practical uses include finding datasets related to a specific tumor type or experimental approach, identifying computational tools for data analysis, locating publications linked to a dataset of interest, discovering researchers working on related problems at other institutions, and accessing educational materials and tutorials. 

Recent news releases from the portal illustrate the type of content being regularly added. The portal expanded its consortium coverage in 2024 and 2025 to formally include the Cellular Cancer Biology Imaging Research program, the Metastasis Research Network, the Cancer Tissue Engineering Collaborative, and the Patient-Derived Models of Cancer program as contributors, alongside the original Cancer Systems Biology Consortium and Physical Sciences in Oncology Network. This expansion substantially broadened the scope of available datasets and tools. The portal’s news feed publishes rolling updates on newly deposited items, including new datasets (for example, single-cell RNA sequencing data characterizing tumor microenvironments in metastatic breast cancer and melanoma drug-resistance studies) and new analytical tools (such as spatial transcriptomics software integrating tissue imaging with gene expression data).

Accessing and organizing information from an increasing number of scientific studies poses a major challenge for researchers, especially in fields such as toxicology that rely on structured data from a variety of studies. Manual data extraction from scientific publications is time-consuming and labor-intensive. To address this issue, DOE and NIEHS researchers developed an automated workflow that uses large language models within the KNIME platform to extract structured data from scientific papers and general PDF files (Moreira-Filho et al. 2025). This workflow combines advanced document processing tools with large language models to extract variables from publications. The workflow has two modes: text mode, which extracts structured text and tables from sources such as peer-reviewed journal articles, and image mode, which processes more complex layouts such as nonstandardized reports or graphical content. The workflow achieved 81.14% accuracy in extracting data from scientific papers using text mode and up to 98.54% accuracy in handling general PDF files using image mode. The free and open-source KNIME data analytics platform, designed with user-friendly features, makes these cutting-edge tools accessible to researchers without specialized expertise. By automating data extraction, this approach saves time and effort while ensuring critical toxicological data are organized and ready for use in chemical assessments, study evaluations, and validation of NAMs. The success of this workflow demonstrates how advanced technologies such as large language models can revolutionize biomedical research by making data retrieval faster and more efficient. By democratizing access to these tools, the platform empowers researchers to synthesize knowledge more effectively, driving progress in toxicology and other fields.

Nano environmental health and safety (nanoEHS) is the study of environmental and health effects of nanomaterials. Multiple U.S. federal agencies have generated vast amounts of nanoEHS data, but sharing and organizing this information has been challenging for researchers, regulators, and partners due to differences in terminology and formats among datasets. To address this, EPA has developed two tools. The NaKnowBase is a curated, relational database that stores experimental results generated by the EPA Office of Research and Development from 2012 to 2019 on how nanomaterials interact with biological and environmental systems (Mortensen et al. 2024a; Boyes et al. 2022). The EPA OntoSearcher tool semiautomates the semantic mapping process. Additional training materials for these tools are available on the NaKnowBase tools webpage.

A federal nanoEHS consortium (Mortensen et al. 2024b), formed through collaboration with groups such as the National Nanotechnology Initiative’s Nanotechnology Environmental Health Implications Working Group and the Database and Informatics Interest Group, aims to establish a "common language" for nanoEHS data. This aligns data with FAIR standards to improve their usability while also ensuring compatibility with European and global partners. EPA has incorporated semantic integration into the NaKnowBase, enabling data entries to be processed into standardized formats such as Research Description Frameworks (Mortensen et al. 2022) for better cross-dataset interoperability.

EPA also developed the OntoSearcher tool and corresponding training materials to assist federal partners with semantic mapping of agency-specific, siloed nanoEHS datasets (Mortensen et al. 2025). This reduces the manual effort required to connect nanoEHS data with ontologies (structured vocabularies), saving time and improving accuracy. These advancements support the development of computational tools designed to enhance data accessibility and integration. Through these efforts, EPA is promoting FAIR data standards and creating tools that improve the organization, sharing, and accessibility of nanoEHS data. These developments set the stage for more efficient collaboration, better risk assessments, and expanded research opportunities in nanomaterial safety. U.S. federal partners have increased interoperability across publicly available but siloed datasets, making it possible to explore the utility of machine learning and super intelligence approaches to improve nanoEHS data content (Mortensen, Singh, et al. submitted). 

OECD has developed harmonized templates for standardized reporting of chemical information including potential human health and environmental effects. OECD Harmonised Template (OHT) 201: Intermediate Effects aims to capture mechanistic information from chemical test results generated by NAMs, such as in vitro testing or in silico methods.

NICEATM developed a workflow for curating HTS assay data from EPA’s invitrodb (v. 4.2) that included an annotation scheme to provide biological context to cHTS data and enable toxicological interpretation. The scheme incorporates annotations from the OBO Foundry, a harmonized and interoperable database consisting of multiple knowledge areas to encompass a broader range of biologic and toxicologic processes. These annotations are available through the NIH Integrated Chemical Environment. Assay description documents based on OECD Guidance Document 211, which outlines ways to harmonize nonguideline in vitro methods, have been made available for ToxCast assay endpoints through invitrodb. These documents describe the details of each assay technology, performance metrics, and data analysis procedures, which align with information requested for OHT 201 fields.

Ongoing annotation efforts by NICEATM, EPA, and the European Joint Research Centre include:

• Identification of HTS-relevant OHT 201 fields
• Alignment of OBO Foundry annotations and mapping information from NICE and invitrodb to OHT 201 fields
• Development of automation workflows to upload information to the International Uniform Chemical Information Database

RxNorm is a widely used standardized drug naming system containing over 1.1 million drug names from 14 different sources. However, within this system, the same drug can be referred to by many different names (brand names, generic names, abbreviations), making it difficult for computer programs to accurately tally drug mentions and associated side effects. To address this problem, researchers at FDA NCTR developed an improved database of drug names to better identify adverse drug events from unstructured text sources such as medical literature and reports (Guo et al. 2025). Using RxNorm data, researchers systematically removed problematic entries including duplicates, extremely long names (over 200 characters), very short names that rarely appear in real-world documents, and names that would cause the same drug to be counted multiple times. The refined database contains approximately 102,000 drug names. When tested on 18,000 drug-related scientific abstracts from PubMed, the refined database identified drugs three times faster than the original RxNorm database while also improving accuracy by eliminating over 1,400 false or redundant drug name matches. The test found that most problematic drug names came from single sources rather than being validated across multiple databases, highlighting the importance of cross-validation. This refined drug name database provides a valuable tool for monitoring medication safety through automated analysis of medical literature, patient reports, and other unstructured healthcare data.