Results Data for "AI-Supported Extraction of Functional Tissue Unit Properties for Human Reference Atlas Construction"

Result Overview Results of AI-Supported Extraction of Functional Tissue Unit Properties for Human Reference Atlas Construction Yongxin Kong1,2, * and Katy Börner1,* 1Department of Intelligent Systems Engineering, Luddy School of Informatics, Computing, and Engineering, Indiana University, Bloomington, IN 47408, USA 2Centre for Regenerative Medicine and Health, Hong Kong Institute of Science & Innovation, Chinese Academy of Sciences, Hong Kong, 999077, China The Human Reference Atlas (HRA) effort brings together experts from more than 25 international consortia to capture the multiscale organization of the human body—from large anatomical organ systems (macro) to the single-cell level (micro). Functional tissue units (FTU, meso) in 10 organs have been detailed and 2D illustrations have been created by experts. Comprehensive FTU property characterization is essential for the HRA, but manual review of the vast number of scholarly publications is impractical. Here, we introduce Large-Model Retrieval-Augmented Generation for HRA FTUs (HRAftu-LM-RAG), an AI-driven framework for scalable and automated extraction of FTU-relevant properties from scholarly publications. This validated framework integrates Large Language Models for textual reasoning, Large Vision Models for visual interpretation, and Retrieval Augmented Generation for knowledge grounding, offering a balanced trade-off between accuracy and processing efficiency. We retrieved 244,640 PubMed Central publications containing 1,389,168 figures for 22 FTUs and identified 617,237 figures with microscopy and schematic images. From these images and associated text, we automatically extracted 331,189 scale bars and 1,719,138 biological entity mentions, along with donor metadata such as sex and age. All data and code are freely available at https://github.com/cns-iu/hra-ftu-rag-supporting-information in support of robust and scalable computation of scholarly evidence for the size and morphological properties and their dependence on donor demographics— facilitating the design, review, and approval of future FTU illustrations during HRA construction. Table description Table Name Attribute Description ftu_publication Purpose To map the function tissue units to the PubMed Central (PMC) publications. Fields Column Data type Description ftu string Function tissue unit. pmcid string The PubMed Central identifier of a publication. publication_su mmary Purpose To store a concise summary of key metadata for PMC publications. Fields Column Data type Description pmcid string The PubMed Central identifier of a publication. article_title string Title of the publication. pmid string The PubMed identifier. doi string The Digital Object Identifier. abstract string The abstract text of the publication. pub_year string The year in which the publication was published. journal_title string The name of the journal in which the publication appeared. file_path string PMC filesystem path to the downloaded file for this publication. Relations hips Linked to "ftu_publication" table via "pmcid". image_summar y Purpose To store a summary of each image associated with a PMC publication. Fields Column Data type Description pmcid string The PubMed Central identifier of a publication. figid string The identifier for the figure within the publication. label string The display label of the figure. graphic string The graphic object identifier. caption string The full caption text for the figure. file_path string The PMC filesystem path to the source file. Relations hips Linked to "ftu_publication" table via "pmcid". image_type Purpose To store the image type for each image. Fields Column Data type Description pmcid string The PubMed Central identifier of a publication. graphic string The graphic object identifier. ext string File extension of the image. is_micro string Indicates whether the image is a microscopy image. is_statis string Indicates whether the image is a statistical image. is_schema string Indicates whether the image is a schematics image. is_3d string Indicates whether the image is a 3D structure diagram. is_chem string Indicates whether the image is a chemical structure diagram. is_math string Indicates whether the image is a mathematical expression. Relations hips Linked to "image_summary" table via "pmcid". Linked to "image_summary" table via "graphic". scale_bar Purpose To store information about scale bar measurements associated with images in PMC publications. Fields Column Data type Description ftu string Function tissue unit. pmcid string The PubMed Central identifier of a publication. graphic string The graphic object identifier. value string The numeric length of the scale bar as reported in the figure. unit String The unit associated with the scale bar value. Relations hips Linked to "image_type" table via "pmcid". Linked to "image_type" table via "graphic". donor_metadat a Purpose To store donor metadata extracted from PMC publications. Fields Column Data type Description pmcid string The PubMed Central identifier of a publication. id string Internal identifier within the publication that denotes the location or object containing the donor metadata. type string Context type indicating where the metadata was found. species_tag string Tag for the donor’s species. develop_tag string Developmental stage category for prenatal and postnatal. sex_tag string Sex tag for the donor. bmi_tag string Body-mass index category. age_yearold_tag string Numeric age in years. Relations hips Linked to "publication_summary" table via "pmcid". biological_struc ture Purpose To store biological entities extracted from PMC publications for 22 FTUs, along with contextual information about where each entity appears (e.g., figure captions, figure nodes). Fields Column Data type Description Relations hips pmcid string The PubMed Central identifier of a publication. id string An internal identifier within a single article that denotes the location or object containing the entity. type string The context type indicating where the entity was found (i.e., “caption”, “figure_node”). Entity string The extracted biological term or name. Relations hips Linked to "publciation_summary" table via "pmcid".