{"id":18185,"date":"2026-08-13T13:38:37","date_gmt":"2026-08-13T17:38:37","guid":{"rendered":"https:\/\/info.hsls.pitt.edu\/updatereport\/?p=18185"},"modified":"2026-09-01T09:00:27","modified_gmt":"2026-09-01T13:00:27","slug":"ai-native-tools-to-support-scholarly-communication","status":"publish","type":"post","link":"https:\/\/info.hsls.pitt.edu\/updatereport\/ai-native-tools-to-support-scholarly-communication\/","title":{"rendered":"AI-Native Tools to Support Scholarly Communication"},"content":{"rendered":"

HSLS has developed five AI-native tools<\/a> to help researchers find evidence, understand content, assess manuscripts, verify citations, and plan scientific visuals. Built with Claude Sonnet 5 through the University of Pittsburgh\u2019s licensed Claude for Education platform. \u00a0The tools are available to Pitt faculty, staff, and students through Pitt Passport single sign-on.<\/p>\n

Research Query Answering Engine<\/strong><\/h3>\n

Accepts research questions written in plain English and after searching and reviewing relevant literature, provides evidence-based answers. Claude expands each question using related terminology and searches sources such as PubMed, PubMed Central, ClinicalTrials.gov, the FDA, and the World Health Organization. References are verified before they are displayed, helping prevent fabricated or hallucinated citations from appearing in the report.<\/p>\n

For example, one could ask, \u201cDoes IL-17 blockade affect cardiovascular risk in patients with psoriasis?\u201d and receive a structured answer supported by verified references, along with an evidence summary, confidence rating, and discussion of limitations. The tool supports rapid exploration of a topic but does not replace a systematic or librarian-assisted literature search. <\/p>\n

PaperSummarizer<\/strong><\/h3>\n

Creates summaries tailored to different audiences. After uploading a paper, users select a research field, an audience\u2014General Audience, Clinician-Scientists, or Expert Researchers\u2014and a desired reading level.<\/p>\n

A faculty member could create a plain-language summary for a community presentation and a technical version for a journal club. A clinician-scientist could request an emphasis on effect sizes, safety signals, and clinical implications. Users should verify all statistics and conclusions against the original paper.<\/p>\n

Research Paper Quality Assessor<\/strong><\/h3>\n

Produces a structured, peer-review-style assessment of a life sciences paper. It identifies the study type, applies a relevant reporting guideline\u2014such as PRISMA, CONSORT, or STROBE\u2014and evaluates the study design, methods, statistics, reproducibility, interpretation, ethics, strengths, and concerns.<\/p>\n

Authors can use the report to identify unclear methods or unsupported conclusions. It does not replace formal peer review or subject-matter, statistical, and methodological expertise. Confidential peer-review and editorial materials must not be uploaded.<\/p>\n

Citation Integrity Audit AI<\/strong><\/h3>\n

First checks whether references appear to exist by comparing bibliographic information with PubMed, Crossref, and other web sources. It then examines claim-citation pairs using available PubMed abstracts to determine whether the cited articles support the manuscript\u2019s statements. Results can be downloaded as CSV files for manual review.<\/p>\n

Authors can use it to identify mistyped DOIs, incorrect publication details, unverifiable references, and citations that may not adequately support a claim.<\/p>\n

PaperSketch Planner<\/strong><\/h3>\n

Converts a paper or pasted description into a plan for a scientific visual. It creates a ready-to-use prompt for image-generation tools such as Google Gemini or Microsoft 365 Copilot, along with a BioRender construction guide containing icon search terms, layout suggestions, and color palettes for manually creating scientific figures.<\/p>\n

A trainee could use it to plan a graphic for a conference poster, while a grant writer could develop a schematic illustrating a proposed mechanism or workflow. The tool provides a visual plan rather than a publication-ready figure.<\/p>\n

Important Limitations<\/strong><\/h3>\n

All five tools are beta releases and can make mistakes. Outputs must be independently verified before they are used in manuscripts, grants, teaching materials, or clinical decisions.<\/p>\n