bims-librar Biomed News
on Biomedical librarianship
Issue of 2026–10–25
forty-nine papers selected by
Thomas Krichel, Open Library Society
  1. Sound solutions: expanding access through digital audiobooks in an academic health sciences library.
  2. Advancing library roles in research analytics: a case study of data-driven grant proposal support service.
  3. Academic librarians as navigators of institutional data policies: developing resources to facilitate compliant research data management.
  4. Still a filtering failure? A follow-up commentary on automated MeSH indexing.
  5. Serials Spoken Here.
  6. Removing records on animals and nonhumans in Ovid Embase: a comparison of 11 filters.
  7. Development and validation of a geographic search filter for PubMed for retrieving studies about Nordic countries.
  8. The state of library publishing in the health sciences in the United States and Canada.
  9. Author profiles in Web of Science, Scopus, Dimensions, ORCID, ResearchGate, and Google Scholar: a case report.
  10. A focused PubMed artificial intelligence search program creates better literature reviews than general purpose artificial intelligence and is comparable to trainees in shoulder and elbow surgery.
  11. Head to head searching: the Search&Destroy card game in the health sciences classroom.
  12. Health sciences librarians' awareness and incorporation of informed consent standards for medical image publication: a preliminary study.
  13. Erratum to: Lane TA, Albro M, Forbes CS. Prevalence of bullying in academic medical libraries in the United States. J Med Libr Assoc. 2026 Jul;114(3):246-254. DOI: 10.5195/jmla.2026.2244.
  14. Beyond database training: reconceptualizing information literacy and reading competence as a shared cognitive architecture in medical education.
  15. Supporting researcher compliance: assessing institutional readiness for the 2023 NIH Data Management and Sharing Policy.
  16. AI, libraries, and literacy: revisiting the Journal of the Medical Library Association's policy on the use of generative artificial intelligence in submissions.
  17. Comparing AI chat tools and PubMed for medical literature searches.
  18. Causal inference methods for systematic reviews.
  19. The All of Us dataset in scientific research: a bibliometric study of its utilization through 2024.
  20. Readability and Citation Sources of ChatGPT-Generated Responses to Lymphoma-Related Health Questions: Cross-Sectional Study.
  21. Evidence-based design of vaccine hesitancy information resources for health care providers.
  22. Comparison of ChatGPT, Gemini, and DeepSeek responses to questions on medication-related osteonecrosis of the jaw (MRONJ): a comparative analysis of expert-rated clinical quality, readability, response time, and clinical safety risk.
  23. How well do large language models answer postoperative lumbar fusion questions? A blinded comparative analysis of accuracy, quality, readability, and safety-related content.
  24. Generative AI chatbots as sources of phimosis-related consumer health information: a cross-sectional comparison of quality, source transparency, and readability.
  25. Performance of large language models in providing patient- oriented information on gingival recession: a comparative study.
  26. Performance of AI chatbots in responding to patients' frequently asked questions about laminate veneer restorations.
  27. Artificial Intelligence as a Digital Consultant: Evaluating Large Language Models for Patient Inquiries on Dental Caries.
  28. Large language models as sources of diabetic retinopathy patient education: a comparative evaluation of safety, health information quality, and readability.
  29. Readability, reliability, and quality of AI-Generated responses on pelvic congestion syndrome: A comparative analysis of ChatGPT, gemini, and perplexity.
  30. Evaluation of large language models for ICI-related liver injury with an emerging microbiome focus.
  31. Information quality, understandability, actionability and readability of large language models for HIV health education: A cross-platform evaluation.
  32. How AI agents search.
  33. Beyond Accuracy: Do Large Language Models Provide Clinically Adequate Responses to Parent-Initiated Questions on Pediatric Obsessive-Compulsive Disorder?
  34. Bridging the Literacy Gap in Breast Reconstruction Patient Education Using Artificial Intelligence.
  35. Why Do Young Adults Turn to YouTube for Diet and Exercise Information? A Comprehensive Model of Information Seeking Approach.
  36. Decision-relevant patient information in YouTube videos on robotic surgery for endometriosis: a cross-sectional content analysis.
  37. Analysis of educational quality of minimally invasive pancreaticoduodenectomy videos on YouTube: a cross-sectional study.
  38. Percutaneous nephrolithotomy on YouTube: Valuable educational resource for urologists and patients or misleading information?
  39. Are parents receiving reliable information on nocturnal enuresis from YouTube?
  40. Quality and reliability evaluation of type 1 diabetes mellitus-related videos on social video platforms: A cross-sectional study.
  41. Gut Health Claims and Monetization Features on TikTok: Cross-Sectional Analysis.
  42. Cross-sectional evaluation of TikTok video quality on alopecia areata.
  43. Content, Engagement, and Information Quality of Borderline Personality Disorder Videos on TikTok: Mixed Methods Analysis.
  44. Quality, reliability, and user engagement of hepatitis B-related short videos on TikTok, Bilibili, and Xiaohongshu: a cross-platform cross-sectional study.
  45. Evaluating influenza-related short videos on Chinese platforms: information quality, user engagement, and public health implications.
  46. "A Minefield for the Novice": Australian Women's Experiences Seeking Health Information Online About Perimenopause.
  47. Pharmacist resource use across the United States: implications for teaching and practice.
  48. Accelerating metadata annotation in collaborative research centers: a hybrid AI workflow for biomedical entities.
  49. Scholarly publishing literacy and perceptions of predatory publishing among educational partners at a chiropractic college: a case report.
  1. J Med Libr Assoc. 2026 Oct 01. 114(4): 439-444
    Sound solutions: expanding access through digital audiobooks in an academic health sciences library.
    Julie Evener, Melinda H Berg.
       Background: As digital audiobooks gain popularity among the general public, academic health sciences libraries have the opportunity to provide access to supplemental nonfiction titles in this format which while often found in public libraries has not been traditionally offered in this setting.
    Case Presentation: An academic health sciences library serving students and faculty in the colleges of medicine, nursing, pharmacy, and public health piloted a collection of nonfiction health sciences audiobooks through OverDrive's Libby app. We launched an initial collection of over 500 titles using the cost per circ (CPC) pricing model, including a focus on four content areas: popular nonfiction, first person experiences, history of healthcare, and study helps. Over the first year of the project, we saw 123 total checkouts of 64 titles across 74 unique users, spending $351.84 US on CPC checkouts.
    Conclusions: The OverDrive title library includes sufficient offerings for academic health sciences libraries to create targeted nonfiction audiobook collections. Challenges of which other libraries should be aware include planning budgets around adding funds each year and determining local workflows to keep the collection updated. Though usage was low in the first year, it was consistent and varied, making the project beneficial and sustainable.
    Keywords:  Audiobooks; Libby; OverDrive; academic health sciences libraries
    DOI:  https://doi.org/10.5195/jmla.2026.2425
  2. J Med Libr Assoc. 2026 Oct 01. 114(4): 483-490
    Advancing library roles in research analytics: a case study of data-driven grant proposal support service.
    Fei Yu.
       Background: This case report demonstrates how a health sciences librarian designed and delivered a data-driven research analytics service to support a multi-institutional pre-award proposal.
    Case Presentation: The librarian was invited to join a multi-institutional research team. After defining the project scope, the librarian developed and executed a structured search strategy in a key research database for collaborative research publications across the participating institutions. After two reviewers independently screened search results for relevance, the librarian applied validated measures and metrics and conducted quantitative data analysis and visualization with bibliographic metadata collected from the included research publications. The findings offered quantitative evidence that illustrates the dynamics and patterns in research productivity, collaboration networks, and topic distribution within the collaborative research effort. This librarian's work contributes to a timely delivery of actionable insights, directly informing the research team's proposal strategy and preparation.
    Conclusion: This case presents a replicable framework for embedding research analytics services within library practice to support research performance evaluation. Leveraging librarians' expertise in literature search, research databases, bibliometrics, and data-driven workflows, librarians demonstrate their growing prominence as essential partners in research proposal development and research analytics.
    Keywords:  Pre-award support; bibliometrics; research analytics; research collaboration
    DOI:  https://doi.org/10.5195/jmla.2026.2412
  3. J Med Libr Assoc. 2026 Oct 01. 114(4): 477-482
    Academic librarians as navigators of institutional data policies: developing resources to facilitate compliant research data management.
    Michelle Yee, Nicole Contaxis.
       Background: Proper research data management requires complying with data policies that ensure the ethical, effective, and efficient collection, storage, and sharing of research data. Medical librarians have made increasing inroads to leading research data management instruction and institutional efforts to comply with federal policies like the NIH Data Management and Sharing Policy. However, there has been less discussion about how medical librarians can support researcher compliance with institutional data policies. After observing patterns in requests for guidance, two data services librarians developed decision-making tools to help researchers identify appropriate local resources for their data sharing and re-use needs.
    Case Presentation: The paper describes four projects that were developed with a similar methodology for the purpose of guiding patrons through institutional policies and processes: (1) a decision tree for determining when a data transfer or use agreement is required; (2) a matching tool that identifies appropriate data transfer platforms for specific data characteristics or scenarios; (3) a decision tree for determining whether a researcher needs to seek out Institutional Review Board (IRB) review for a project that uses secondary data analysis; and (4) a flowchart detailing data de-identification processes and institutional tools available for specific data formats. The creation of each resource entailed information gathering from local experts as well as review of internal policies to produce action-oriented guidance to solve common researcher problems.
    Conclusions: These efforts demonstrate how librarians can utilize their skills and develop cross-departmental collaborations to help patrons navigate an increasingly complicated data ecosystem, particularly where that ecosystem is informed both by external policies (e.g., policies for de-identification) and internal tools (e.g., data transfer tools). The decision trees have helped researchers make choices about their own data management approaches so that they comply with external and internal policies. These documents also enable stakeholders to identify gaps in institutional policies and resources. The development of these resources has had the additional benefit of strengthening partnerships and positioning librarians as collaborators for policy decisions involving research data.
    Keywords:  Data management; academic librarians; data policy; educational guidance; federal regulation
    DOI:  https://doi.org/10.5195/jmla.2026.2404
  4. J Med Libr Assoc. 2026 Oct 01. 114(4): 508-511
    Still a filtering failure? A follow-up commentary on automated MeSH indexing.
    Nicole Askin, Tyler Ostapyk, Carla Epp.
      We previously found that automated indexing in PubMed and Medline increased the risk of inappropriately excluding human studies when using the human studies filter from the Cochrane Highly Sensitive Search Strategy (HSSS). Following the National Library of Medicine's transition from the Medical Text Indexer-Auto (MTIA) to the Medical Text Indexer-Next Generation (MTIX) in 2024, we wanted to assess whether that was still the case. Using the same approach as our previous study, we reviewed a set of MTIX-indexed records that were excluded by the HSSS. Only 1.1% turned out to be human studies that were inappropriately excluded - a major improvement over not only MTIA but also human indexers in our previous comparison. However, problems remain with automated indexing and search filters. This commentary outlines our findings and our ongoing concerns.
    Keywords:  Automated MeSH indexing; commentary; filter
    DOI:  https://doi.org/10.5195/jmla.2026.2377
  5. Ser Rev. 2026 Sep 29.
    Serials Spoken Here.
    Michelle Colquitt.
      This edition of "Serials Spoken Here" contains a wide variety of session recaps. Conference sessions from both national/international conferences as well as state/local sessions. Recaps from the ALA Annual Convention & Exhibition, Electronic Resources & Libraries (ER&L), UKSG, the Acquisitions Institute at Timberline Lodge, the American Association of Law Libraries, NISO Plus, ExLibris Users of North America (ELUNA), and the Academic Libraries of Indiana (ALI) Resource Rendezvous. Topics are arranged under the broad headings of Collection Management, Contracts & Agreements, Legal & Government Serials, Scholarly Integrity, Workflows, Metrics, and Discovery. The editors hope you will take something away from this article that could potentially impact your current workflows or spark ideas for further research. We would also like to encourage you to contribute to the column by sharing recaps of interesting conference sessions you attend.
    Keywords:  COUNTER; Serials management; metrics; transformative agreements; workflows
    DOI:  https://doi.org/10.1080/00987913.2026.2729465
  6. J Med Libr Assoc. 2026 Oct 01. 114(4): 380-392
    Removing records on animals and nonhumans in Ovid Embase: a comparison of 11 filters.
    Helen Fulbright, Connor Evans.
       Objective: To understand and compare the impact of 11 filters to remove records on animals or nonhumans in Ovid Embase, which are routinely used despite there being no performance data for them. To understand the indexing of relevant subject headings in Embase.
    Methods: To assess filter performance, we screened and categorized 3,000 records as 'should be removed' or 'should be retained' and calculated the sensitivity, specificity, and overall accuracy for each filter. We also reported on the number of records with human participants or data that were incorrectly removed by each filter.
    Results: Method 11 was the most sensitive, correctly retaining 90.6% records, whereas method 3 had the highest specificity, correctly removing 71.5% records. Of the records incorrectly removed by each filter, method 11 removed the fewest records with human participants or data at 10.9%, whereas method 3 removed the most at 14.3%. Filters that did not remove records indexed with 'nonhuman/' had higher sensitivity. Filter performance could vary by subject, publication type and language due to differences in indexing.
    Conclusion: In choosing a search filter, information specialists and review teams should discuss whether animals or nonhumans could feature in relevant studies.
    Keywords:  Bibliographic database; Embase; MEDLINE; indexing; information retrieval; search strategies
    DOI:  https://doi.org/10.5195/jmla.2026.2437
  7. J Med Libr Assoc. 2026 Oct 01. 114(4): 419-427
    Development and validation of a geographic search filter for PubMed for retrieving studies about Nordic countries.
    Malin Barkelind.
       Objective: This study describes the development and validation of a geographic search filter. The filter is intended to find references on the Nordic countries in PubMed. This type of filter enables efficient systematic literature searches for evidence that is relevant for the region, and no validated filter for these countries is publicly available at the time of this publication.
    Methods: The Nordic countries are Denmark, Finland, Iceland, Norway and Sweden. References on one or more of these countries were identified by relative recall, using systematic and scoping reviews that did not limit searches to any geographic region but presented the countries of included studies. The references (n = 288) were equally split into a development set and a validation set. A combination of objective and subjective methods was used to identify relevant terms from the development set. The filter was then tested, first on the development set, then on the validation set. A sample of the reviews that were used to identify references for recall testing was also used to test precision.
    Results: The final filter surpassed the threshold recall of 90%. Recall was 94.4% for the development set, capturing 136 of 144 references, and 96.5% for the validation set, or 139 of 144. Precision was 95.8% with one false positive of 24 retrieved references.
    Conclusions: A validated filter has been developed that can be used to find studies on the Nordic countries when searching PubMed on topics related to human health.
    Keywords:  PubMed; Scandinavian and Nordic Countries; information storage and retrieval
    DOI:  https://doi.org/10.5195/jmla.2026.2311
  8. J Med Libr Assoc. 2026 Oct 01. 114(4): 428-438
    The state of library publishing in the health sciences in the United States and Canada.
    Gina Genova.
       Objective: Publishing initiatives led by academic libraries have grown over the past decade. They are often involved in the development of innovative publishing models and open initiatives, and a number of library publishers have worked in the health sciences. This study aims to assess the current state of library publishing in the health sciences (HS).
    Methods: The Library Publishing Directory survey and Library Publishing Coalition's member directory were accessed to identify active library publishing programs in the U.S. and Canada. Survey responses on disciplinary specialties and top publications were assessed, and individual publisher websites were accessed to inventory HS titles. Journals were checked for whether they were actively publishing and indexed.
    Results: One hundred eighteen survey respondents and 24 additional LPC members were assessed. Twenty-five percent of respondents named an HS subject as a disciplinary specialty, and 20% named an HS title as a top publication. Two hundred seventy-five individual HS titles from 81 different library publishers were identified. Fifty-three percent were journals, 39% open education resources, and 8% monographs. Medicine was the most common field for publications. Seventy-three percent of journals were actively publishing. Of those, 42% were indexed in DOAJ or a major HS database, and just over 10% appeared in PubMed or Scopus.
    Conclusions: Library publishers have a track record of working with a wide variety of HS researchers. Indexing of library-published journals is an area for improvement. HS librarians may be able to collaborate with their institutions' publishing programs to further develop these services and build relationships with researchers.
    Keywords:  Library publishing; scholarly communication
    DOI:  https://doi.org/10.5195/jmla.2026.2278
  9. J Med Libr Assoc. 2026 Oct 01. 114(4): 491-498
    Author profiles in Web of Science, Scopus, Dimensions, ORCID, ResearchGate, and Google Scholar: a case report.
    Kimberly R Powell.
       Background: This study aimed to benchmark the availability and accuracy of author profiles for 45 faculty members within the Auburn University College of Nursing. Profile availability in the database platforms Web of Science, Scopus, and Dimensions were analyzed alongside the author-driven platforms ORCID, ResearchGate, and Google Scholar to set an understanding for the potential efforts needed for profile curation, publication tracking, and research impact assessments moving forward.
    Case Presentation: Each of the six platforms were manually searched for faculty and name variations alongside both current and past institutional affiliations. All identified profiles were collected. Associated publication counts were recorded and database results were screened for accurate author identification. Among the author-driven platforms, ORCID had the highest adoption rate, with 64% of faculty, while ResearchGate and Google Scholar trailed with 38% and 16% faculty adoption, respectively. Web of Science offered profiles for 82% of faculty; Scopus offered profiles for 80% of faculty, and Dimensions only 53%. However, 53% of the associated publications were misattributed in Web of Science profiles compared to just 1% misattribution in Scopus and none in Dimensions.
    Conclusions: Among our faculty, Scopus offered the most efficient author profile platform, balancing profile availability, minimizing misattribution, and including avenues for early identification of new authors. While largely in line with previous study findings, this effort provided important contextual understandings of profile data availability which can be used to inform research impact assessments and library services supporting author profiles for highlighting scholarly influence.
    Keywords:  Dimensions; Google Scholar; ORCID; ResearchGate; Scopus; Web of Science; author disambiguation; author identifiers; researcher impact
    DOI:  https://doi.org/10.5195/jmla.2026.2304
  10. JSES Int. 2026 Nov;10(6): 101790
    A focused PubMed artificial intelligence search program creates better literature reviews than general purpose artificial intelligence and is comparable to trainees in shoulder and elbow surgery.
    James S MacLeod, Michael S Lee, Hayden Baker, Jordan Serotte, Lewis L Shi, Nicholas H Maassen.
       Background: As the use of artificial intelligence (AI) expands into orthopedic surgery, the potential to use AI for review and summation of literature leaves researchers and clinicians questioning the quality of AI-driven literature reviews. This study aims to compare AI-generated literature reviews to those done by orthopedic surgery trainees.
    Methods: ChatGPT-4o and The-Literature, a directed AI platform, were asked to produce a literature review on 5 shoulder and elbow-related research topics. The primary output produced by ChatGPT-4, The-Literature, a fourth-year medical student, a third-year orthopedic surgery resident, and an orthopedic surgery sports medicine fellow were blinded then graded on a 1-5 Likert Scale for evidence and analysis, conclusion, references, and overall quality by 2 fellowship-trained orthopedic shoulder and elbow surgeons.
    Results: The average overall Likert ratings were 2.90 ± 0.88 for the medical student, 3.90 ± 0.74 for the resident, 3.50 ± 0.53 for the fellow, 2.30 ± 1.16 for ChatGPT-4, and 3.40 ± 0.84 for The-Literature. The resident and fellow reviews were significantly better than ChatGPT-4 (P = .0004, P = .0209), while the medical student review was not (P = .5092). None of the trainee reviews were significantly better than The-Literature. The-Literature's reviews were significantly better than ChatGPT-4 (P = .0412). The majority (58.8%) of ChatGPT-4's references were fabricated.
    Conclusion: A limited and directed AI search program was comparable to orthopedic surgery trainees and performed better than an open AI large language model. ChatGPT-4 had poor overall utility stemming mainly from its tendency to fabricate references.
    Keywords:  AI; Artificial intelligence; ChatGPT; Elbow; Literature review; Shoulder; The-literature
    DOI:  https://doi.org/10.1016/j.jseint.2026.101790
  11. J Med Libr Assoc. 2026 Oct 01. 114(4): 462-470
    Head to head searching: the Search&Destroy card game in the health sciences classroom.
    Lena Bohman, Regina Vitiello, Christiana M Keinath, Gary R Maixner, Mari Kermit-Canfield.
       Background: Games are a promising way to make health sciences education more engaging and improve synthesis for learners. There have been some successful implementations of game-based learning for health sciences library instruction [1], but there is room for development of games tailored specifically for teaching database searching skills to health sciences students.
    Case Presentation: The authors developed a biomedical edition of the existing library instruction card game Search&Destroy. The original Search&Destroy was developed as an alternative way to teach a one-shot instructional session on search skills using a discovery layer in an academic library for undergraduates. The biomedical edition modified the game to work with PubMed and incorporate health science-specific searching skills. Keyword cards were updated with biomedical terms and new cards were added with filters and actions related to health sciences literature searching (e.g., last five years, systematic reviews, randomized controlled trials filters). The new version was play tested internally by the authors and then piloted in classes with nurse residents. The game is now available through the print-on-demand service The Game Crafter and teaching materials are available online.
    Conclusions: Feedback from our pilot test group of nurse residents was overwhelmingly positive. Informal conversations with nurse educators responsible for the program were also positive, and the game has been incorporated in the curriculum for the future. The Biomedical Edition joins a growing family of Search&Destroy games designed for a variety of library instruction needs.
    Keywords:  Game-based learning; card games; game design; game development; games; gamification
    DOI:  https://doi.org/10.5195/jmla.2026.2299
  12. J Med Libr Assoc. 2026 Oct 01. 114(4): 401-409
    Health sciences librarians' awareness and incorporation of informed consent standards for medical image publication: a preliminary study.
    Sarah McClung, Rachel Keiko Stark, Megan DeArmond, Sherli Koshy-Chenthittayil.
       Objective: The objective of this study was to determine if health sciences library workers were aware of informed consent best practices and standards for the publication of medical images and if they incorporated the recommendations into their professional work.
    Methods: A survey was developed by the authors to determine awareness and incorporation of informed consent best practices and standards for the publication of medical images. The survey was distributed via listservs of professional librarian associations in the United States. The results of the survey were tabulated in RStudio. Fisher's Exact Test was carried out to determine if awareness of informed consent influenced job responsibilities.
    Results: A total of 91 respondents were included in the data analysis with all respondents reporting multiple responsibilities (instructional, collections, and/or digital initiatives). While the majority of library workers (59%) were aware of the best practices and standards, few (6.6%) incorporated the recommendations into their professional work. Over 50% of librarians with instructional responsibilities, over 70% of librarians with collections responsibilities, and over 60% of librarians with digital initiative responsibilities were aware of the recommendations but did not incorporate the recommendations into their job duties.
    Conclusions: The professional work of health sciences library workers does not appear to include a significant incorporation of informed consent best practices and standards for the publication of medical images. There is a need for future research to better understand how library workers can better incorporate their knowledge of informed consent recommendations into their work.
    Keywords:  Consent; health sciences librarians; images; publication
    DOI:  https://doi.org/10.5195/jmla.2026.1949
  13. J Med Libr Assoc. 2026 Oct 01. 114(4): E1
    Erratum to: Lane TA, Albro M, Forbes CS. Prevalence of bullying in academic medical libraries in the United States. J Med Libr Assoc. 2026 Jul;114(3):246-254. DOI: 10.5195/jmla.2026.2244.
    Todd Lane.
      [This corrects the article DOI: 10.5195/jmla.2026.2244.].
    DOI:  https://doi.org/10.5195/jmla.2026.2612
  14. J Med Libr Assoc. 2026 Oct 01. 114(4): 499-507
    Beyond database training: reconceptualizing information literacy and reading competence as a shared cognitive architecture in medical education.
    Mercedes Byrd, James K Martin, Alexis Karas.
      Medical librarians have long observed that medical students who struggle with reading complex textbooks also demonstrate weak information literacy skills. For decades, these have been treated as separate problems requiring separate solutions-offering reading workshops and database training sessions with limited lasting impact. This commentary proposes a reconceptualization: textbook literacy and information literacy are not sequential competencies but rather parallel manifestations of the same underlying cognitive architecture. Drawing on research in medical education, cognitive psychology, and educational neuroscience, this commentary argues that both literacies depend on three interconnected cognitive capacities: metacognitive self-regulation, working memory management, and epistemic sophistication. Understanding this shared foundation has profound implications for how health sciences librarians design interventions, plan instruction, and position themselves as essential contributors to medical education. This commentary presents a framework and outlines practical implications for library instruction, assessment, and collaboration with medical educators.
    Keywords:  Information literacy; cognitive load theory; epistemic development; health sciences librarianship; medical education; metacognition
    DOI:  https://doi.org/10.5195/jmla.2026.2385
  15. J Med Libr Assoc. 2026 Oct 01. 114(4): 371-379
    Supporting researcher compliance: assessing institutional readiness for the 2023 NIH Data Management and Sharing Policy.
    Sergio Corrales-Guerrero, Xu Zhang, Thipkanok Wongphothiphan, Jenny Xu, Yungui Huang.
       Objective: To assess researchers' understanding, readiness, and perceived challenges related to the 2023 NIH Data Management and Sharing (DMS) Policy and identify institutional support needs to facilitate compliance.
    Methods: A mixed-method study was conducted using a 15-item survey distributed to 155 researchers at an academic medical center, with 55 responses (35% response rate). The survey evaluated familiarity with the DMS policy, confidence in performing data management and sharing tasks, perceived burden, and desired institutional support. Quantitative data were analyzed descriptively, and open-ended responses underwent inductive thematic analysis to identify key themes.
    Results: Most respondents (67%) understood the DMS policy, yet fewer (47%) felt comfortable developing a DMS Plan. Only 29% had used DMPTool.org, while 87% requested tailored templates. Confidence in identifying appropriate repositories (40%) and budgeting for DMS activities (38%) was limited, with 68% perceiving plan execution as burdensome. Qualitative findings revealed two major themes: (1) Operational and Infrastructure Needs; emphasizing the need for centralized consultation, discipline-specific templates, equitable access to expertise, and budget guidance; and (2) Implementation Challenges and Perceived Burden; including perceptions of administrative overload, uncertainty in managing sensitive data, and calls for clearer institutional communication and cultural reframing of data sharing as a value-added activity.
    Conclusions: While researchers endorse the principles of the NIH DMS policy, many lack the practical resources and institutional support necessary for effective compliance. Developing centralized infrastructure, standardized guidance, and targeted training may reduce burdens, promote equity, and strengthen a culture of responsible data stewardship. While researchers endorse the principles of the NIH DMS policy, many lack the practical resources and institutional support necessary for effective compliance. Developing centralized infrastructure, standardized guidance, and targeted training may reduce burdens, promote equity, and strengthen a culture of responsible data stewardship.
    Keywords:  Data Sharing Compliance; Institutional Support; NIH Data Management and Sharing Policy; Policy Implementation; Research Data Stewardship; Researcher Perceptions
    DOI:  https://doi.org/10.5195/jmla.2026.2361
  16. J Med Libr Assoc. 2026 10 01. 114(4): 330-335
    AI, libraries, and literacy: revisiting the Journal of the Medical Library Association's policy on the use of generative artificial intelligence in submissions.
    Alexander J Carroll, Jill T Boruff, Michelle Kraft.
      In the fall of 2023, the Journal of the Medical Library Association (JMLA) introduced its policy on generative artificial intelligence (AI) submissions, predicting that these tools would become embedded within the workflows of clinicians, educators, researchers, and students. Those predictions dramatically underestimated AI's impact. On the three-year anniversary of that editorial, we revisit the issue in light of AI's unprecedentedly rapid adoption across knowledge work. We examine the educational challenges AI poses, arguing that excessive cognitive offloading threatens the deep thinking on which learning depends. We then consider the challenges AI creates for researchers, from hallucinated citations to fabricated diseases. We suggest that health sciences librarians, who offer expertise in approaching information systems with discernment, are well suited to be leaders in this space. To help prospective JMLA authors weigh whether and when to use AI, we propose the Compliance, Verifiability, and Disclosure (CVD) Framework. Throughout, we reaffirm our conviction that education and research must remain human-centered endeavors.
    Keywords:  AI in Libraries; AI literacy; Artificial Intelligence; information literacy; lifelong learning; literacy skills; scholarly communication; scholarly publishing; scholarly writing
    DOI:  https://doi.org/10.5195/jmla.2026.2516
  17. J Med Libr Assoc. 2026 Oct 01. 114(4): 359-370
    Comparing AI chat tools and PubMed for medical literature searches.
    A'Llyn Ettien, Kathleen Downing, Rebecca Drouhard, Bridget Gunn, Megan McNichol, Kate Silfen, Allie Tatarian.
       Objective: This investigation compared six AI chat tools (ChatGPT, Claude, Copilot, Gemini, OpenEvidence, and Perplexity) to each other and to PubMed to assess their utility in medical literature searches.
    Methods: Each tool was prompted with two standardized prompts simulating an academic and a clinical literature need. Prompts were repeated three times per tool, and results recorded for each prompt iteration.
    Results: Main findings include the range in number of real and hallucinated articles for each AI tool, from no hallucinations in OpenEvidence to more than 80% hallucinated citations in ChatGPT. Across all tools, most real articles were assessed as relevant. Continued prompting of all tools produced duplicate results. Most tools showed very little consistency in citations provided to different researchers, suggesting low probability of replicating search results.
    Conclusions: Though limited by small sample size, inconsistent data, and the rapid pace of change in LLM technology, this study suggests some advantages and drawbacks of using AI tools for literature searching compared with PubMed. AI tools with few or no hallucinations and high relevance, such as OpenEvidence and Perplexity, performed well in terms of simply supplying a few article citations on the topic, but small results sets and lack of consistency and transparency limit their value for thorough research. These results suggest that AI tools are not presently a viable replacement for PubMed or other academic databases for systematic reviews or other in-depth literature searches but may be adequate for limited information needs or as a complement to database or web searches.
    Keywords:  AI; comparison; literature searches
    DOI:  https://doi.org/10.5195/jmla.2026.2284
  18. J Med Libr Assoc. 2026 Oct 01. 114(4): 336-342
    Causal inference methods for systematic reviews.
    Madelyn Glymour.
      Systematic reviews are frequently used to ask and answer causal questions, and decisions made throughout the review process can introduce bias because of the causal relationships between variables of interest, even when the question being asked is not itself causal. However, current systematic review best practices do not consider most of these impacts. The author presents an introduction to the basic principles and terminology of causal inference as apply to systematic reviews and discusses how some simple causal inference concepts can be used to create clearer and less biased research questions, search strategies, eligibility criteria, and data analysis. By taking into account confounder and collider bias, and by paying careful attention to the differences in assumptions made in intervention- and non-intervention-based reviews, health science librarians can aid systematic review teams in minimizing the risk of bias, and in appropriately analyzing and assessing that risk.
    Keywords:  Systematic review; causal inference; collider bias; confounder bias; methodology
    DOI:  https://doi.org/10.5195/jmla.2026.2253
  19. J Med Libr Assoc. 2026 Oct 01. 114(4): 343-358
    The All of Us dataset in scientific research: a bibliometric study of its utilization through 2024.
    Yuqi He, Abrielle Spaletta.
       Objective: This study presents a comprehensive bibliometric analysis of research publications utilizing the All of Us Research Program dataset to characterize publication trends, collaboration networks, key research topics, and compliance with All of Us publication and presentation policies through December 2024.
    Methods: A total of 561 publication titles were collected from the All of Us Publications Directory. Titles were then searched across Web of Science (WoS), Scopus, and PubMed, with WoS (475 publications identified) selected as the primary source for analysis. Productivity metrics, including publication growth, journal distribution, author productivity, and word frequency, were analyzed using Biblioshiny. Co-authorship and keyword co-occurrence networks were visualized in VOSviewer to reveal collaboration structures and emerging research themes.
    Results: Among the 475 analyzed publications, 2,803 authors were identified across 211 journals. The annual number of All of Us publications increased by 82.45% from 2015 to 2024. Major research themes included epidemiology, chronic conditions, genomics, and health disparities, with dermatology emerging as a particularly active area. Network analysis revealed strong intra-institutional ties and some cross-institutional collaborations, primarily among R1 universities.
    Conclusions: This bibliometric study demonstrates a rapid rise in scholarly activity leveraging the All of Us dataset, characterized by concentrated author and institutional engagement. These findings highlight both the growing impact of the All of Us dataset and opportunities to foster broader collaborations and diversify participation across the research community.
    Keywords:  All of Us; Biblioshiny; VOSviewer; bibliometric analysis; collaboration network; dataset utilization; policy compliance; publication trend
    DOI:  https://doi.org/10.5195/jmla.2026.2346
  20. Online J Public Health Inform. 2026 Oct 23. 18 e95067
    Readability and Citation Sources of ChatGPT-Generated Responses to Lymphoma-Related Health Questions: Cross-Sectional Study.
    Corey H Basch, Tailisha González, Grace C Hillyer, Erin T Jacques, Charles E Basch.
       BACKGROUND: Lymphoma is one of the most prevalent groups of cancers among people aged 15 to 29 years. Increasingly, patients with lymphoma and their caregivers use generative AI systems, supported by large language models (LLMs), to seek cancer-related online health information. Despite the need for plain and clear health information for clinical decision-making, there is a dearth of research on the readability and citation source classification of LLM-supported online health information.
    OBJECTIVE: The objective of this study was to assess the reading level and reading ease of lymphoma-related health information using OpenAI's GPT-5.2 model, as well as the citation sourcing of LLM-generated results.
    METHODS: Questions were submitted to the LLM through an automated pipeline. The application code was written with Claude and run and hosted on Vercel. Standardized instructions were submitted to the LLM, asking for evidence-based sources with URLs to enable citation analysis. The final dataset included 392 citations garnered from 96 standardized lymphoma-related questions. Readability was assessed using the Flesch-Kincaid Grade Level and Flesch Reading Ease scores. Citation sources were classified into 9 predefined organization types, and topics included Hodgkin and non-Hodgkin lymphoma and clinical trials. Differences in readability were tested using linear models fitted to all 392 citations, with standard errors clustered by question. Estimated marginal means (EMMs) were used to summarize model-adjusted readability scores by citation source and topic. When the omnibus test was statistically significant, pairwise comparisons were performed using the Tukey-Kramer adjustment for multiple comparisons.
    RESULTS: The Flesch-Kincaid reading levels of the ChatGPT responses by citation source and topic type were determined to be at the college or university level, well above the recommended sixth-grade level. Grade level did not differ by citation source (P=.58). The reading grade level of responses for Hodgkin lymphoma responses was statistically higher than that of clinical trial responses (mean 16.2, SD 3.4 vs 13.6, SD 2.5; P=.005). Flesch Reading Ease scores across source and topic were found to be very difficult at the university graduate academic level and did not differ by source (P=.27) or topic (P=.17).
    CONCLUSIONS: Although LLMs frequently produce responses that include cited information, the readability of these outputs may limit accessibility for many patients and caregivers. These findings highlight the need for improved readability standards in AI-generated health information and reinforce that LLM outputs should complement, not replace, professional medical guidance.
    Keywords:  AI; artificial intelligence; health information–seeking behaviors; large language models; lymphoma; online health information; plain and clear language
    DOI:  https://doi.org/10.2196/95067
  21. J Med Libr Assoc. 2026 Oct 01. 114(4): 452-461
    Evidence-based design of vaccine hesitancy information resources for health care providers.
    Margaret Ansell, Hannah F Norton, Nina C Stoyan-Rosenzweig, Jane Morgan-Daniel, Mary E Edwards, Lauren Adkins.
       Background: The University of Florida Health Science Center Libraries (HSCL) sought to address the increasing challenge of vaccine hesitancy during the COVID-19 pandemic by developing evidence-based information resources for health care providers. To do this, HSCL librarians assessed the vaccine hesitancy information needs of local health care providers to design a resource that could be used to support conversations with vaccine hesitant patients at the point of care.
    Case Presentation: A team of six librarians, supported by student research assistants, conducted interviews with 20 health care providers in north central Florida. The interviews, conducted via Zoom, focused on providers' experiences with vaccine-hesitant patients. Content analysis of the transcribed interviews identified key themes, including trust, individualized conversations, and the impact of misinformation. The team developed a resource guide based on these findings, featuring curated information to support providers in vaccine discussions.
    Conclusions: The results highlighted the importance of trust in patient-provider relationships and the challenges posed by misinformation. Providers expressed a need for diverse, accessible resources to aid in conversations about vaccines. The resulting resource guide, disseminated through local health networks, aims to enhance providers' ability to address vaccine hesitancy effectively. This project underscores the critical role of librarians in public health initiatives and the value of collaborative efforts to combat vaccine hesitancy.
    Keywords:  Vaccine hesitancy; clinical information resources; evidence-based librarianship; qualitative data analysis
    DOI:  https://doi.org/10.5195/jmla.2026.2135
  22. BMC Oral Health. 2026 Jul 25. pii: 2072. [Epub ahead of print]26(1):
    Comparison of ChatGPT, Gemini, and DeepSeek responses to questions on medication-related osteonecrosis of the jaw (MRONJ): a comparative analysis of expert-rated clinical quality, readability, response time, and clinical safety risk.
    Emre Ulubaş, Damla Maide Davutluoğlu, İnci Rana Karaca, Mehmet Emin Toprak.
       BACKGROUND: Medication-related osteonecrosis of the jaw (MRONJ) is a clinically significant complication associated with antiresorptive and antiangiogenic therapies, and its prevention and management depend on reliable, guideline-concordant information. Large language models (LLMs) are increasingly used to generate medical information, but their performance in MRONJ remains insufficiently investigated. This study compared the responses generated by ChatGPT- 5.4 Thinking, Gemini 3 Deep Think, and DeepSeek-V3.2 in thinking mode to MRONJ-related open-ended questions in terms of clinical quality, readability, and response time.
    METHODS: In this cross-sectional comparative study, a standardized 30-item question set was developed from the MASCC/ISOO/ASCO Clinical Practice Guideline and the Italian Consensus Update. Each question was entered separately into each model under standardized conditions, yielding 90 responses. Outputs were anonymized and independently evaluated by three blinded oral and maxillofacial surgeons using a modified Global Quality Scale on a five-point Likert scale. A post hoc Clinical Risk Classification was additionally performed to distinguish minor informational deficiencies from omissions or inaccuracies with greater potential clinical consequences. Readability was assessed using five established English-language indices, and response time was measured with an online digital stopwatch. Appropriate statistical tests were applied, and inter-evaluator reliability was assessed using the intraclass correlation coefficient.
    RESULTS: Inter-evaluator agreement was high across all models (all p < 0.001). Mean expert quality scores differed significantly among models (p < 0.001). Post hoc comparisons showed that ChatGPT-5.4 Thinking received significantly lower expert ratings than both DeepSeek-V3.2 and Gemini 3 Deep Think, whereas no statistically significant difference was detected between DeepSeek-V3.2 and Gemini 3 Deep Think. Response time also differed significantly (p < 0.001): DeepSeek-V3.2 produced the fastest responses, Gemini 3 Deep Think showed intermediate latency, and ChatGPT-5.4 Thinking had the longest response times. Readability analyses revealed significant inter-model differences across all indices, with Gemini 3 Deep Think generating the most complex outputs and DeepSeek-V3.2 producing comparatively more readable responses. Clinical Risk Classification showed that most responses were categorized as having no clinically relevant risk or low clinical risk, while moderate-risk classifications were uncommon and no high-risk responses were identified.
    CONCLUSIONS: Under single-prompt, single-session benchmark conditions, potentially clinically relevant differences were observed among the models in MRONJ-related response quality, readability, response time, and clinical risk classification. DeepSeek-V3.2 and Gemini 3 Deep Think achieved higher expert-rated quality scores than ChatGPT-5.4 Thinking, while DeepSeek-V3.2 also showed advantages in speed and readability. These findings do not indicate general model superiority and support using LLMs only as adjunctive informational tools, not substitutes for specialist and guideline-based clinical decision-making.
    Keywords:  Artificial Intelligence; Large language models; MRONJ; Readability; Response quality
    DOI:  https://doi.org/10.1186/s12903-026-09421-0
  23. Front Surg. 2026 ;13 1932133
    How well do large language models answer postoperative lumbar fusion questions? A blinded comparative analysis of accuracy, quality, readability, and safety-related content.
    Ahmet Kürşat Kara, Ozan Işık.
       Background: Patients recovering from lumbar fusion increasingly seek guidance from large language model (LLM) chatbots when their surgeon is unavailable, but the accuracy, completeness, and safety-related content of such responses have not been compared across models for the postoperative period.
    Objective: To compare the accuracy, quality, safety-related content, and readability of responses from ChatGPT, Gemini, and Claude to frequently asked postoperative lumbar fusion questions, and to identify the question types for which responses are weakest.
    Methods: Thirty questions, compiled by three neurosurgeons and grouped into six clinical categories, were submitted to each model, each with a standardized, patient-oriented prompt requesting warning signs. Ninety anonymized, randomized responses were rated independently by three blinded neurosurgeons for accuracy and four quality subdomains (clarity, completeness, relevance, and patient safety) on 4-point scales. Readability (Flesch-Kincaid, Flesch Reading Ease, and Gunning Fog), word count, and responses meeting an operational screening criterion for potential harm were also assessed. Models were compared with Friedman and Holm-corrected pairwise tests.
    Results: Accuracy was high for all models without significant differences. Quality, clarity, completeness, relevance, and patient safety differed globally, but effect sizes were small (Kendall's W: 0.10-0.21) and, after correction across the six outcome families, only quality and completeness remained significant, with Gemini scoring lower than ChatGPT and Claude. Gemini had the most favorable readability. ChatGPT responses were the shortest (mean 294 vs. 462 and 445 words); length was unrelated to completeness or readability. Two Gemini responses (6.7%) and none from ChatGPT or Claude met a post hoc operational screening criterion for potential harm (a score of ≤2 on accuracy or patient safety from at least one of three raters, without concurrence from the other two); no response was adjudicated as actually harmful. In exploratory question- and category-level analyses, responses tended to be rated highest for standardized questions and lowest for those requiring individualized judgment or interpretation of wound-related symptoms.
    Conclusion: Responses from contemporary LLMs to common postoperative lumbar fusion questions were generally rated as accurate, with small between-model differences and a model-level divergence between readability and completeness. Because safety-related content was explicitly prompted and no patient outcomes were assessed, these findings describe the quality of safety-related information rather than clinical safety. LLMs may support postoperative patient education but should not replace physician guidance, particularly for warning signs and patient-specific recovery decisions.
    Keywords:  large language models; lumbar fusion; patient safety; postoperative patient education; readability
    DOI:  https://doi.org/10.3389/fsurg.2026.1932133
  24. Front Public Health. 2026 ;14 1983152
    Generative AI chatbots as sources of phimosis-related consumer health information: a cross-sectional comparison of quality, source transparency, and readability.
    Qinghua Luo, Leihua Cao, Pan Ding.
       Background: Publicly accessible generative artificial intelligence chatbot products are increasingly used for sensitive health questions, but the quality, source transparency, and readability of phimosis-related information remain uncertain.
    Methods: We conducted an internet-based cross-sectional comparative study on 1 July 2026. 15 of 25 Google Trends candidate terms were submitted unchanged to ChatGPT (displayed as GPT-5.5), Microsoft Copilot, Gemini (displayed as 3.5 Flash), and Perplexity, yielding 60 responses. Two practising urologists holding senior professional titles, each with more than 10 years of clinical experience, independently assessed responses using DISCERN, Ensuring Quality Information for Patients (EQIP), the Global Quality Score (GQS), and Journal of the American Medical Association (JAMA) benchmark criteria. Inter-rater reliability was assessed using ICC (2,1) for DISCERN and EQIP and quadratic-weighted Cohen's kappa for JAMA and GQS. Because the same 15 queries were evaluated across all four products, quality outcomes were analysed as matched data using Friedman tests with Kendall's W and Holm-adjusted paired Wilcoxon signed-rank comparisons. Readability was assessed with six established indices.
    Results: Inter-rater reliability was good to almost perfect (coefficients 0.866-0.882). After Holm adjustment across the four omnibus tests, differences remained for DISCERN (p = 0.007; W = 0.301), EQIP (p < 0.001; W = 0.587), JAMA (p < 0.001; W = 0.953), and GQS (p = 0.049; W = 0.174). For DISCERN, Copilot scored higher than ChatGPT and Gemini, and Perplexity scored higher than Gemini. For EQIP and JAMA, Copilot and Perplexity scored higher than ChatGPT and Gemini. No GQS pairwise comparison remained significant after Holm correction. None of the 15 responses from any product met any individual prespecified readability threshold. A sensitivity analysis excluding the broad standalone prompts 'circumcision' and 'balanitis' preserved the DISCERN, EQIP, and JAMA findings but not the GQS omnibus difference.
    Conclusion: In this purposively derived 15-query, single-date snapshot, chatbot products differed in selected measures of information quality and source transparency, while readability remained consistently above the prespecified sixth-grade target. These findings are conditional on the sampled queries and outputs and should not be interpreted as stable platform-wide estimates of clinical accuracy or safety.
    Keywords:  artificial intelligence; chatbot; consumer health information; health literacy; phimosis; readability; urology
    DOI:  https://doi.org/10.3389/fpubh.2026.1983152
  25. Saudi Dent J. 2026 Oct 19. pii: 138. [Epub ahead of print]38(11):
    Performance of large language models in providing patient- oriented information on gingival recession: a comparative study.
    Sultan Albeshri.
      Patients are increasingly turning to large language models (LLMs) for health information, but the quality of patient-oriented information remains uncertain. This article evaluated the performance of three major LLMs in answering clinically relevant patient's questions related to gingival recession. A descriptive comparative study was carried out utilizing twenty standardized patient-centered questions related to gingival recession. Each question was submitted once to ChatGPT 5.5, Gemini 3.5, and Claude Sonnet 5, generating 60 responses. Five board-certified periodontists independently evaluated the responses using the validated Accuracy of Information (AOI) Index, a completeness scale, and expert preference rankings. Word count and response generation time were also recorded. Non-parametric tests and generalized estimating equations (GEE) were applied for statistical analysis, with statistical significance set at p ≤ 0.05. No significant differences were found among the three LLMs in accuracy (p = 0.993) or completeness (p = 0.972). ChatGPT generated responses significantly faster (median: 5.82 s) and with fewer words (median: 232) than Gemini (8.46 s; 438 words) and Claude (14.61 s; 333.5 words) (both p < 0.001). Expert preference rankings differed significantly (p = 0.002): Gemini and Claude were each ranked first in 38% of evaluations, compared with 24% for ChatGPT. For all three models, the highest accuracy was observed in Post-treatment Care. The three LLMs offered broadly comparable levels of accuracy and completeness when answering patient-oriented questions with respect to gingival recession. ChatGPT was faster and more concise, while Gemini and Claude were more often preferred by the expert evaluators. These LLMs may be useful as adjuncts to patient education, but they should complement rather than replace professional periodontal consultation.
    Keywords:  Artificial intelligence; Chatbots; Dentistry; Gingival recession; Large language models
    DOI:  https://doi.org/10.1007/s44445-026-00247-8
  26. BMC Oral Health. 2026 Jul 28. pii: 2085. [Epub ahead of print]26(1):
    Performance of AI chatbots in responding to patients' frequently asked questions about laminate veneer restorations.
    Bike Altan, Şevki Çinar.
       BACKGROUND: Although AI (artificial intelligence) chatbot systems are increasingly used to obtain dental information, limited evidence exists regarding the reliability and readability of their responses in laminate veneer-related patient inquiries. The aim of the study is to assess the reliability, quality, and readability of the responses given by 3 different artificial intelligence chatbot systems to frequently asked questions about laminate veneer restorations by patients.
    METHODS: Twenty-five frequently asked questions about laminate veneer restorations submitted to ChatGPT 5.2, Gemini 3, and DeepSeek-V3.2. AI chatbots to generate answers. The quality of these answers was assessed by two different experts using GQS (Global Quality Scale), reliability using modified DISCERN, and readability using FRES (Flesch Reading Ease Score) and FKGL (Flesch-Kincaid Grade Level). Normality was assessed with the Shapiro-Wilk test. Depending on data distribution, repeated-measures ANOVA or the Friedman test was used for comparisons among chatbots. Dunn's test was applied for post hoc analyses. A p-value < 0.05 was considered statistically significant.
    RESULTS: No statistically significant differences were found among chatbot systems in terms of DISCERN, FRES, or FKGL scores (P > 0.05). Interobserver agreement for DISCERN was generally poor, with only Gemini demonstrating statistically significant agreement (ICC = 0.393; P = 0.023). Although a significant overall difference was observed in GQS scores (P = 0.015), no significant pairwise differences were detected. Readability analysis revealed that all chatbot responses exceeded the recommended 6th-8th grade reading level for patient education materials, with mean FKGL values corresponding to approximately 1-2 grade levels above the recommended upper limit. In addition, FRES values indicated that the generated content generally corresponded to a difficult reading level.
    CONCLUSION: The evaluated AI chatbot systems demonstrated largely comparable performance in responding to laminate veneer-related inquiries. While overall content quality was high, limitations in reliability and elevated readability levels indicate that AI-generated responses should be interpreted cautiously and used as supportive tools rather than substitutes for professional clinical guidance.
    Keywords:  Artificial intelligence; Laminate veneer; Quality; Readability; Reliability
    DOI:  https://doi.org/10.1186/s12903-026-09438-5
  27. J Esthet Restor Dent. 2026 Oct 22.
    Artificial Intelligence as a Digital Consultant: Evaluating Large Language Models for Patient Inquiries on Dental Caries.
    Betül Erdal, Fatmanur Merve Duman, Mehmet Robin Altınkılıç, Meltem Horoz.
       OBJECTIVE: To evaluate the clinical accuracy, usefulness, global overall quality (GQS), and readability of ChatGPT, Gemini, and DeepSeek responses to patient questions regarding dental caries.
    MATERIALS AND METHODS: Fifty-nine simulated patient questions concerning caries diagnosis, etiology, treatment, and prevention were queried in Turkish. Three restorative dentists independently evaluated the responses for accuracy, usefulness, and GQS. Readability was assessed using the Ateşman index. Statistical analyses included the intraclass correlation coefficient (ICC) for interrater reliability, the Friedman test for performance comparisons among the models, and Spearman's rank correlation to analyze the relationships between the clinical variables.
    RESULTS: Inter-rater reliability was high across all criteria (ICC > 0.75). Significant performance differences existed among the models (p < 0.001); Gemini achieved the highest overall clinical accuracy, usefulness, and GQS scores. ChatGPT produced the most readable texts, categorized generally as "easy to read" according to the Ateşman index. For DeepSeek, higher clinical accuracy significantly correlated with better readability (p < 0.001).
    CONCLUSION: Gemini offers the highest accuracy and quality as a digital preconsultation tool. However, AI tools should complement, not replace, professional dental advice.
    Keywords:  ChatGPT; DeepSeek; Gemini; artificial intelligence; dental caries; health literacy
    DOI:  https://doi.org/10.1111/jerd.70284
  28. Front Public Health. 2026 ;14 1945318
    Large language models as sources of diabetic retinopathy patient education: a comparative evaluation of safety, health information quality, and readability.
    Zhiqi Chen, Wen Liu, Jing Zhang, Tian Zheng, Ying Li, Qing Bao, Pingting Ji, Hongmei Yao, Ying He.
       Background: Patients increasingly use large language models (LLMs) to seek medical information, but their performance in answering diabetic retinopathy (DR) questions remains insufficiently characterized.
    Objective: To evaluate the safety, accuracy, patient-centered communication, information quality, transparency, and readability of diabetic retinopathy information generated by widely accessible large language models.
    Methods: Fifty standardized DR patient questions were submitted to ChatGPT, Gemini, DeepSeek, Doubao, Copilot, and Perplexity. Responses were independently evaluated by five raters using guideline-based reference answer elements. Rater-based outcomes included Safety, Accuracy, Empathy, DISCERN, EQIP, JAMA benchmark criteria, and Global Quality Score (GQS). Readability was assessed using the Automated Readability Index, Coleman-Liau Index, Flesch-Kincaid Grade Level, Gunning Fog Index, SMOG, and Flesch Reading Ease Score. Inter-rater agreement and between-model differences were analyzed using appropriate agreement statistics and paired nonparametric tests.
    Results: Inter-rater agreement was moderate for Safety (Fleiss' κ = 0.581) and good for non-binary outcomes, with ICC (2,1) values ranging from 0.806 to 0.870. Most responses were classified as safe, with safe response rates ranging from 86.0 to 94.0%, and no significant between-model difference was observed. Accuracy scores were generally high across models, although between-model differences were statistically significant with a small effect size. Empathy varied more noticeably, with Copilot scoring highest and Doubao and Perplexity scoring lower. Perplexity achieved the highest DISCERN and JAMA scores, while Copilot performed best on EQIP. GQS differences were small. Readability differed significantly across models, with Gemini generally producing the most readable responses; however, all models exceeded the recommended sixth-grade reading level, and all Flesch Reading Ease Scores were below the recommended threshold.
    Conclusion: No single LLM performed best across all domains. Although the evaluated models generally provided safe and accurate responses to DR patient questions, limitations remained in safety escalation, empathy, transparency, and readability. LLM-generated DR information may support general patient education but should not replace professional ophthalmic consultation.
    Keywords:  diabetic retinopathy; digital health (eHealth); health communication (MESH); health literacy; large language models (LLM); patient education; patient safety-standards
    DOI:  https://doi.org/10.3389/fpubh.2026.1945318
  29. Phlebology. 2026 Oct 18. 2683555261499766
    Readability, reliability, and quality of AI-Generated responses on pelvic congestion syndrome: A comparative analysis of ChatGPT, gemini, and perplexity.
    Emre Sertel.
      ObjectivePelvic congestion syndrome (PCS) is a vascular cause of chronic pelvic pain arising from ovarian and pelvic venous insufficiency, typically presenting as dull, aching pelvic pain that worsens with prolonged standing, physical activity, and the premenstrual period, and is particularly common among multiparous women of reproductive age. Online health information on this condition is frequently accessed by both patients and healthcare professionals. This study aimed to evaluate the readability, quality, and reliability of responses generated by three popular artificial intelligence (AI) chatbots-ChatGPT-4o, Gemini, and Perplexity-regarding the most commonly asked questions on PCS.MethodsFrom 25 PCS-related keywords identified through Google Trends, duplicates and irrelevant terms were excluded, yielding a final set of 18 keywords. These were separately submitted to the three AI chatbots. Readability was assessed using the Gunning Fog Index (GFOG), Simple Measure of Gobbledygook (SMOG), and Flesch Reading Ease Score (FRES), among other indices, calculated via two independent readability tools. Quality was measured using the Global Quality Score (GQS) and the Ensuring Quality Information for Patients (EQIP) tool, while reliability was assessed with the Journal of the American Medical Association (JAMA) benchmark criteria and the modified DISCERN instrument.ResultsReadability analyses revealed that all chatbot responses significantly exceeded the sixth-grade level recommended by the AMA and NIH (p < .001). ChatGPT-4o achieved the highest readability on the CLI and ARI, whereas Gemini performed better on the remaining indices, and Perplexity consistently produced the most complex responses. Among the three evaluated freely accessible platforms, Perplexity achieved the highest scores on several quality and reliability measures (JAMA, DISCERN, and GQS), while Gemini outperformed the others on EQIP scoring.ConclusionAlthough AI chatbots hold promise for providing accessible information on PCS, their readability levels remain above recommended standards, and overall quality scores are suboptimal. Producing clearer, more balanced, and more reliable content could enhance the contribution of these technologies to public health.
    Keywords:  ChatGPT; artificial intelligence; gemini; online medical information; pelvic congestion syndrome; perplexity
    DOI:  https://doi.org/10.1177/02683555261499766
  30. Front Cell Infect Microbiol. 2026 ;16 1955356
    Evaluation of large language models for ICI-related liver injury with an emerging microbiome focus.
    Man Sun, Dan Zang, Huan Zhou, Jun Chen.
       Background: Immune checkpoint inhibitor-related liver injury (ICI-LI) has heterogeneous phenotypes, a complex differential diagnosis, and evolving management. Gut microbiome research provides biological plausibility for immune regulation during ICI therapy, but evidence specific to ICI-LI remains limited and is not yet clinically actionable. We evaluated whether publicly accessible large language model (LLM) chatbots can organize this mixed evidence base and communicate its uncertainty.
    Methods: We developed a disease-specific benchmark of 25 questions covering immunological mechanisms, microbiome and immune regulation, diagnosis, biomarkers, and management of ICI-LI. The benchmark addressed ICI-LI broadly and included a focused microbiome component. Four publicly available LLM platforms were evaluated using DISCERN, EQIP, the Global Quality Score, the JAMA Benchmark Criteria, and six readability indices. These instruments assessed information quality, structure, transparency, and patient-facing readability rather than factual accuracy or clinical decision-making.
    Results: Performance differed across DISCERN, EQIP, and JAMA scores, whereas the Global Quality Score showed limited discrimination. Question-level variation was most apparent for items requiring evidence appraisal or translation. All platforms produced structured responses, but none met the prespecified patient-facing readability benchmark. Readability was interpreted separately from clinician-facing information synthesis.
    Conclusions: Disease-specific multidimensional evaluation is needed before LLM-generated information can be considered reliable for microbiome-informed clinical communication or decision support. Evaluation should emphasize evidence calibration and distinguish biological plausibility, association, and clinically actionable evidence.
    Keywords:  gut microbiome; gut–liver axis; immune checkpoint inhibitor-related liver injury; large language models; precision medicine
    DOI:  https://doi.org/10.3389/fcimb.2026.1955356
  31. HIV Med. 2026 Oct 18.
    Information quality, understandability, actionability and readability of large language models for HIV health education: A cross-platform evaluation.
    Mengqiu Hu, Zhiqiang Wang, Jinglan Zhang, Xiaoya Li, Mengxia Zhao.
       OBJECTIVES: Large language models (LLMs) are increasingly used to generate health education materials, but their suitability for patient-facing human immunodeficiency virus (HIV) education remains insufficiently evaluated. We assessed the information quality, understandability, actionability, readability, text structure and source transparency of HIV patient education materials generated by six LLMs or artificial intelligence platforms.
    METHODS: In this cross-sectional study, 100 common HIV patient education questions across 10 domains were submitted once in English to ChatGPT 5.4-Thinking, DeepSeek-V3.2, Doubao, ERNIE Bot 5.0 Preview, Ant Afu and Gemini 3.1 Pro, yielding 600 responses. Responses were evaluated using Patient Education Materials Assessment Tool for Printable Materials (PEMAT-P), EQ IP-36, DISCERN, Global Quality Score (GQS), artificial intelligence response quality metric (AIRM), Journal of the American Medical Association (JAMA) benchmarks and seven readability indices. Platform comparisons used paired Friedman tests with Wilcoxon-Holm post hoc analyses; topic analyses used question-level aggregation followed by Kruskal-Wallis and Dunn-Holm comparisons.
    RESULTS: Significant family-adjusted differences were observed across platforms for all prespecified text-structure, quality/understandability/actionability and readability outcomes. At the topic level, five of six text-structure outcomes, PEMAT-P Understandability, PEMAT-P Actionability, the exploratory PEMAT-P composite score and all seven readability indices differed significantly after family adjustment, whereas EQ IP-36, DISCERN, GQS, AIRM and JAMA did not. Overall actionability was limited and readability demands were high.
    CONCLUSIONS: LLM-generated HIV patient education materials showed substantial cross-platform and topic-level heterogeneity. These findings support the use of LLMs as expert-reviewed drafting tools rather than independent patient education sources. Direct factual accuracy, guideline concordance, hallucination and clinical safety were not assessed.
    Keywords:  HIV; actionability; health education; information quality; large language models; patient education; readability
    DOI:  https://doi.org/10.1111/hiv.70329
  32. PLoS One. 2026 ;21(10): e0360157
    How AI agents search.
    Francesca Pezzuti, Ophir Frieder, Fabrizio Silvestri, Sean MacAvaney, Nicola Tonellotto.
      Large Language Models (LLMs) have evolved from relying primarily on knowledge stored in their parametric memory to using non-parametric memory mechanisms such as Retrieval-Augmented Generation (RAG) to provide and ground knowledge. Recently, LLMs have begun to perform retrieval autonomously within a reasoning loop, an approach referred to as Agentic RAG. This shift introduces new querying behaviours, with implications for the design of retrieval systems that have yet to be evaluated. In this work, we introduce a large-scale dataset (ASQ) capturing how Agentic RAG systems perform multi-step search and reasoning. ASQ contains about 1M agentic search traces collected from open-source Agentic RAG systems over 3 corpora, 5 agent variants, and 6 retrieval settings. Our analysis of ASQ shows that, within this controlled benchmark setting, a single search query can trigger long chains of retrieval calls (up to 186) and follow query reformulation dynamics different from those observed in historical human-search data. Our results also reveal substantial behavioural variation across agent and retrieval configurations: changes in re-ranking and retrieval depth do not consistently affect search workloads in terms of search-call volume and trace length across agents. To our knowledge, publicly available production logs do not expose the agent, dataset, and retrieval configuration associated with each trace. ASQ provides a reusable resource for further analysing agent search behaviour and supporting future information retrieval studies on agentic workloads under controlled settings.
    DOI:  https://doi.org/10.1371/journal.pone.0360157
  33. Clin Child Psychol Psychiatry. 2026 Oct 22. 13591045261499579
    Beyond Accuracy: Do Large Language Models Provide Clinically Adequate Responses to Parent-Initiated Questions on Pediatric Obsessive-Compulsive Disorder?
    Berrin Bilgiç, Serkan Turan.
      BackgroundParents increasingly use large language models (LLMs) to obtain information about child and adolescent mental health. In pediatric obsessive-compulsive disorder (OCD), where early recognition and intervention are critical, the accuracy of AI-generated information may have clinical implications. This study evaluated LLM responses to parent-oriented questions about pediatric OCD.MethodsAn exploratory cross-sectional evaluation was conducted using three conversational AI systems: ChatGPT (GPT-4o), Google Gemini, and DeepSeek. Twenty questions reflecting common parental concerns were identified through online sources and expert review and submitted with standardized prompts in separate chat sessions. Anonymized responses were independently rated by two child and adolescent psychiatrists for accuracy, reproducibility, and error patterns.ResultsChatGPT achieved the highest accuracy (≈91%) and reproducibility (≈89%), followed by Gemini (≈88%; ≈85%) and DeepSeek (≈86%; ≈83%). Domain-level analysis showed the highest scores for early recognition/diagnosis and treatment-related questions, and the lowest scores for questions on family functioning and daily life. Omission of clinically critical information-particularly insufficient emphasis on exposure and response prevention-was more common than factual inaccuracy.ConclusionsLLMs can provide accessible and generally accurate information for parents about pediatric OCD. However, variability across domains and limited prioritization of clinically essential content suggest they should complement, not replace, professional evaluation.
    Keywords:  child and adolescent psychiatry; health information seeking; large language models; obsessive-compulsive disorder
    DOI:  https://doi.org/10.1177/13591045261499579
  34. Aesthetic Plast Surg. 2026 Oct 23.
    Bridging the Literacy Gap in Breast Reconstruction Patient Education Using Artificial Intelligence.
    Erin Kim, Ethan D Paliwoda, Rhett Farmer, Melia Wakeman, Hanaa Shihadeh, Dzifa Kpodzo, Benjamin J Schalet.
       BACKGROUND: Artificial intelligence is increasingly integrated into healthcare, demonstrating applications across surgical prediction, intraoperative precision, and patient communication. However, there is limited literature examining the role of artificial intelligence in improving patient comprehension of educational materials for breast reconstruction.
    METHODS: 30 texts from online education resources for breast reconstruction surgery were selected. Readability of the texts was then evaluated using the Flesch-Kincaid grade level, the Gunning Fog index, and the Coleman-Liau index. ChatGPT 4.0 was instructed to translate the individual texts to the 6th-grade level of English comprehension, and the readability of the translated texts was analyzed. Patient Education Materials Assessment Tool analysis assessed the understandability and actionability of each text.
    RESULTS: Shapiro-Wilk test on the paired differences showed that the Flesch-Kincaid grade level and the Gunning Fog index followed normal distributions, whereas the Coleman-Liau index did not. The mean Flesch-Kincaid grade level, the Gunning Fog index, and the Coleman-Liau index for the original texts were 11.2, 13.9, and 11.3. The mean Flesch-Kincaid grade level, the Gunning Fog index, and the Coleman-Liau index for the AI-translated counterparts were 8.6, 10.5, and 9.7 (all p < 0.001). The mean understandability (0.70 SD = 0.06 vs. 0.58 SD = 0.08, p < 0.001) and actionability (0.40, SD = 0.09 vs. 0.37 SD = 0.11, p = 0.057) scores of the artificial intelligence-translations were higher than that of the original texts upon Patient Education Materials Assessment Tool analysis.
    CONCLUSION: Artificial intelligence may be a promising tool for enhancing comprehension of breast reconstruction education materials among individuals with limited health literacy.
    LEVEL OF EVIDENCE IV: This journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors www.springer.com/00266 .
    Keywords:  Artificial intelligence; Health literacy; Patient education; Plastic surgery; Readability
    DOI:  https://doi.org/10.1007/s00266-026-06400-2
  35. Health Commun. 2026 Oct 17. 1-16
    Why Do Young Adults Turn to YouTube for Diet and Exercise Information? A Comprehensive Model of Information Seeking Approach.
    Eunyoung Myung, Karyn Riddle.
      This study examines why U.S. young adults (18-29) frequently rely on YouTube for diet and exercise information, drawing on the comprehensive model of information seeking (CMIS) as a guiding framework. Using a national survey of 508 respondents, we found that health consciousness, internet self-efficacy, convenience motivations, perceived credibility, and perceived vividness significantly predict perceived utility of YouTube. In turn, perceived utility mediates the relationship between these key antecedents and health information-seeking behaviors. These findings underscore the importance of platform-specific factors, particularly vividness and convenience in shaping health communication strategies aimed at young adults. From a theoretical standpoint, the results expand CMIS by highlighting how interactive, user-generated media environments can amplify or alter traditional predictors of information-seeking. Practically, they suggest that designing YouTube content to be both visually engaging and trustworthy may enhance user perceptions of utility and ultimately encourage healthier behaviors. Overall, this study sheds light on how young adults navigate digital resources and offers valuable insights for improving online health interventions.
    DOI:  https://doi.org/10.1080/10410236.2026.2748522
  36. J Robot Surg. 2026 Oct 22. pii: 1122. [Epub ahead of print]20(1):
    Decision-relevant patient information in YouTube videos on robotic surgery for endometriosis: a cross-sectional content analysis.
    Sema Karakas, Zinar Serhanoglu, Basri Çakıroğlu.
      To evaluate decision-relevant content completeness, patient-oriented communication, and source transparency of YouTube videos on robotic surgery for endometriosis. Cross-sectional content analysis. Publicly accessible YouTube content identified through six English- and Turkish-language search strings in September 2026; the retained records document search dates for only a subset. Seventy-six fully accessible videos explicitly addressing robotic surgery for endometriosis, selected from 101 unique records after exclusion of inaccessible videos and records without confirmed eligibility. No clinical intervention. Two reviewers independently coded eligible videos using the study checklist and resolved differences by consensus; reviewer-specific ratings were not retained. In a post hoc strict recoding of the indication/patient-selection domain from retained video/page notes, 18 videos (23.7%) qualified; a broad surgery/candidacy reference was present in 73 (96.1%). The recalculated median DRCCI was 1 of 9 (IQR, 0-2); under a sensitivity rule counting any broad reference, the median was 2 (IQR, 1-3). Deep endometriosis and organ involvement were discussed in 29 (38.2%) and 25 (32.9%) videos; non-surgical options, comparison with laparoscopy, risks, fertility/ovarian reserve, recurrence, and multidisciplinary care were present in 5 (6.6%), 6 (7.9%), 7 (9.2%), 14 (18.4%), 2 (2.6%), and 5 (6.6%), respectively. Selected decision-relevant topics were infrequently present in this exploratory sample of public videos. Because intended audience and purpose were not classified, these findings do not establish that all videos were intended for patient counseling or that omissions affected viewers.
    Keywords:  Endometriosis; Health communication; Patient education; Robotic surgery; YouTube
    DOI:  https://doi.org/10.1007/s11701-026-04092-z
  37. Ann Surg Treat Res. 2026 Oct;111(4): 185-193
    Analysis of educational quality of minimally invasive pancreaticoduodenectomy videos on YouTube: a cross-sectional study.
    Korean Study Group of Minimally Invasive Pancreatic Surgery (K-MIPS)
       Purpose: Minimally invasive pancreaticoduodenectomy (MIPD) is a complex procedure with a steep learning curve requiring advanced training and surgical proficiency. YouTube (Google LLC) is a widely used platform for surgical education; however, the quality of MIPD videos varies significantly. This study evaluates the educational quality of MIPD videos on YouTube and identifies characteristics of high-quality instructional content.
    Methods: A YouTube search was conducted on November 11, 2024, using the following 4 keywords: "laparoscopic pancreaticoduodenectomy," "laparoscopic Whipple," "robotic pancreaticoduodenectomy," and "robotic Whipple." Among 200 screened videos, 66 met the inclusion criteria. Videos were evaluated using the Laparoscopic Video Educational Guidelines and Standards (LAP-VEGaS) guidelines (0-18) and an MIPD-specific scoring system (0-12).
    Results: The quality of YouTube MIPD videos varied considerably. Although videos from academic societies had more views, they did not consistently have higher scores. The highest-rated videos featured comprehensive elements such as preoperative imaging, anatomical annotations, detailed surgical technique, and structured narration. Top-performing videos achieved scores of up to 16/18 (LAP-VEGaS score) and 12/12 (MIPD-specific score).
    Conclusion: Despite increased reliance on YouTube for surgical education, high-quality MIPD content remains scarce. Academic societies and surgical societies should actively contribute to the production and curation of standardized, high-quality videos to support global surgical training and improve outcomes.
    Keywords:  Medical education; Minimally invasive surgical procedures; Pancreaticoduodenectomy; Social media; Video recording
    DOI:  https://doi.org/10.4174/astr.2026.111.4.185
  38. Cent European J Urol. 2026 ;79(3): 387-392
    Percutaneous nephrolithotomy on YouTube: Valuable educational resource for urologists and patients or misleading information?
    Guglielmo Mantica, Giovanni Drocchi, Rafaela Malinaric, Federica Balzarini, Valentina Catale, Pieter Spies, André Van der Merwe, Daniele Panarello, Francesco Chierigo, Federico Dotta, Matteo Ferro, Carlo Introini, Carlo Terrone.
       Introduction: To evaluate the quality, reliability, and educational value of YouTube videos related to percutaneous nephrolithotomy (PCNL), with a specific focus on the content available for urologists and patients seeking information online.
    Material and methods: A systematic YouTube search using the key words "PCNL" and "percutaneous nephrolithotomy" was conducted. Videos were included if they were in English, had audio, and were relevant to PCNL. Two expert urologists independently evaluated each video. Collected data included upload year, video length, content type, target audience, source, geographic origin, and surgical position. The quality of the videos was assessed using the DISCERN instrument and Global Quality Score (GQS).
    Results: Of 65 identified videos, 26 met inclusion criteria. Most were aimed at healthcare professionals (84.6%) and were uploaded by physicians (65.4%). Only 15.4% were patient-oriented. The prone position was depicted in 84.6% of videos. The median duration was 7 minutes 58 seconds. The median view ratio was 102.3 views/day. The median DISCERN score was 2 (IQR 1), and the median GQS score was 3 (IQR 1.375), reflecting moderate educational quality.
    Conclusions: YouTube contains a moderate-quality body of PCNL videos primarily designed for professional audiences. There is a clear shortage of reliable, patient-centered educational content. Healthcare organizations and scientific societies should proactively develop and promote accessible, high-quality educational resources for patients.
    Keywords:  health information; patient education; social media; surgical education; video-based education
    DOI:  https://doi.org/10.5173/ceju.2025.0240
  39. BMC Urol. 2026 Jul 28. pii: 255. [Epub ahead of print]26(1):
    Are parents receiving reliable information on nocturnal enuresis from YouTube?
    Yunus Nas, Hilmi Bayırlı, Ali Gözüküçük, Basri Çakıroğlu.
       OBJECTIVE: To evaluate the quality, reliability, source characteristics, and content completeness of YouTube videos related to pediatric nocturnal enuresis and to identify potential gaps in online patient education.
    MATERIALS AND METHODS: A cross-sectional analysis of YouTube videos was conducted using the search terms "nocturnal enuresis," "bedwetting," and "child bedwetting." The first 100 eligible videos were included because they represent the content most likely to be encountered by users during routine searches. Video characteristics, including duration, views, likes, comments, uploader source, and content type, were recorded. Educational quality and reliability were assessed using the Global Quality Scale (GQS) and the modified DISCERN instrument. Content completeness was evaluated using a predefined scoring system covering definition, etiology, diagnostic evaluation, treatment options, behavioral interventions, and prognosis.
    RESULTS: A total of 100 videos were analyzed. The median video duration was 3.5 min (IQR: 2.1-9.1), and the median number of views was 1,192 (IQR: 148-4,964). The mean GQS, modified DISCERN, and content completeness scores were 3.22 ± 1.19, 2.69 ± 1.35, and 3.23 ± 1.66, respectively. Significant differences in educational quality, reliability, and content completeness were observed according to uploader source (all p < 0.001). Videos produced by healthcare professionals, academic institutions, and professional organizations generally demonstrated higher quality and more comprehensive content than commercial or non-professional sources. Definitions (78%), causes (74%), and treatment options (81%) were frequently discussed, whereas diagnostic evaluation (49%) and prognosis (38%) were substantially underrepresented. Behavioral interventions were mentioned in 67% of videos.
    CONCLUSIONS: YouTube videos on pediatric nocturnal enuresis demonstrate substantial variability in quality, reliability, and content completeness. Although videos originating from healthcare-related sources tended to provide more comprehensive and reliable information, important gaps remain, particularly regarding diagnostic evaluation and prognosis. These deficiencies may contribute to incomplete parental understanding and highlight the need for evidence-based digital educational resources and clinician-guided online health information.
    Keywords:  Digital health; Health information quality.; Nocturnal enuresis; Patient education; YouTube
    DOI:  https://doi.org/10.1186/s12894-026-02276-1
  40. Health Informatics J. 2026 Oct-Dec;32(4):32(4): 14604582261500294
    Quality and reliability evaluation of type 1 diabetes mellitus-related videos on social video platforms: A cross-sectional study.
    Yuan Hu, Xianrong Li, Jianling Yang, Fengmei Tan, Li Liu, Xinyue Zhou, Xiaolei Zhao.
      ObjectiveThis study aimed to assess the quality and reliability of videos about type 1 diabetes mellitus (T1DM) on TikTok, Bilibili, and YouTube.MethodsThe keyword "type 1 diabetes mellitus" was used to search for related videos on TikTok, Bilibili, and YouTube. The quality and reliability of 263 videos were assessed using the modified DISCERN (mDISCERN), the Global Quality Scale (GQS), and the Journal of the American Medical Association (JAMA) benchmark criteria.ResultsQuality scores of videos on YouTube were higher, and videos uploaded by specialists were rated higher than those posted by non-specialists. However, user engagement metrics (likes, comments, saves, and shares) did not correlate with quality scores, indicating that engagement alone does not reflect content quality.ConclusionThe quality and reliability of videos related to T1DM on these platforms were insufficient. Video platforms should recommend high-quality videos to guide viewers toward reliable sources.
    Keywords:  Quality; Reliability; Type 1 diabetes mellitus; Video
    DOI:  https://doi.org/10.1177/14604582261500294
  41. Online J Public Health Inform. 2026 Oct 22. 18 e105391
    Gut Health Claims and Monetization Features on TikTok: Cross-Sectional Analysis.
    Kai Vu, Kyle Garcia-Rogers, Benjamin Li, Vinayak Swaroop, Owais Gilani, Alice M Tang, Michael Siegel.
       BACKGROUND: TikTok has emerged as a major source of health information, yet concerns persist regarding content accuracy and monetization features. Gut health content is particularly vulnerable to misinformation.
    OBJECTIVE: This study aimed to examine the relationship between creator profession ("medical" vs "nonmedical") and the quality of gut health claims and monetization features on TikTok.
    METHODS: We conducted a cross-sectional study screening 412 unique TikTok creator accounts identified using the keywords "guthealth," "gutcleansing," and "digestion." The final sample included 174 publicly accessible individual creators with at least 5000 followers and identifiable, relevant profession information. One video per creator was analyzed. Creator profession was categorized as medical or nonmedical, and health claim quality was coded as high, moderate, or poor. Monetization features (Showcase, Subscription, and external links) were assessed.
    RESULTS: Nonmedical creators were more likely than medical creators to present poor- or moderate-quality health claims (adjusted prevalence ratio 2.33, 95% CI 1.50-3.62). Most creators (160/174, 92%) exhibited at least 1 monetization feature, and Showcase use was greater among nonmedical creators (adjusted prevalence ratio 1.57, 95% CI 1.02-2.42). Videos containing moderate- and poor-quality health claims received 3 times the expected likes as videos containing high-quality claims.
    CONCLUSIONS: Among the sampled creators, nonmedical creators were more likely to present poor- or moderate-quality gut health claims, and lower-quality claims were associated with higher like counts. These findings highlight a misalignment between information quality and like counts on social media.
    Keywords:  TikTok; gastroenterology; gut health; health misinformation; medical influencers; social media; wellness; wellness influencers
    DOI:  https://doi.org/10.2196/105391
  42. Dermatol Reports. 2026 Oct 19.
    Cross-sectional evaluation of TikTok video quality on alopecia areata.
    Hilla Rosenberg, Chaya Bracha Gordon, Yael Hollander, Reut Rotem, Amir Horev.
      Alopecia areata (AA) is an autoimmune condition that results in hair loss, significantly impacting quality of life. As TikTok emerges as a major platform for health-related information, concerns about the spread of misinformation have arisen. This cross-sectional study aims to evaluate the quality, reliability, and understandability of "most liked" AA-related TikTok videos. Videos were categorized as healthcare providers (HP) or non-healthcare providers (NHP). Two reviewers independently assessed the videos, using validated tools - Patient Education Materials Assessment Tool for audiovisual content (PEMAT A/V), modified DISCERN (mDISCERN), Video Information and Quality Index (VIQI), Global Quality Scale (GQS), and AloCheck, a study-specific content-coverage tool. Among 93 included videos, only 26% were created by HPs, who scored significantly higher across all quality metrics: mDISCERN (3.0 vs. 1.5, p<0.01), GQS (3.5 vs. 1.5, p<0.01), VIQI (13.75 vs. 11.0, p<0.01), and PEMAT A/V (≥10: 66.7% vs. 26.1%, p<0.01). HPs focused on clinical aspects (87.5% vs. 40.6%, p<0.01), whereas NHPs emphasized psychological aspects, such as body image (37.5% vs. 63.8%, p=0.02). Findings reveal a gap between quality and user engagement, highlighting the risk that lower-quality information reaches a wide audience. Bridging this gap is essential to support the delivery of high-quality, reliable, and engaging health information to the public.
    Keywords:  Alopecia areata; adolescents; health education; social media
    DOI:  https://doi.org/10.4081/dr.2026.10952
  43. JMIR Form Res. 2026 Oct 16. 10 e106246
    Content, Engagement, and Information Quality of Borderline Personality Disorder Videos on TikTok: Mixed Methods Analysis.
    Gabrielle Evoy, Simon Poirier, Majorie Vadnais, Alexandre Hudon.
       Background: TikTok has become a major source of mental health information, yet concerns persist regarding the quality and reliability of psychiatric content disseminated on the platform. While misinformation has been documented for several mental disorders, no study has specifically examined how borderline personality disorder (BPD) is portrayed on TikTok.
    Objective: This study aims to characterize TikTok content related to BPD by describing its thematic content, engagement patterns, presenter characteristics, and indicators of information quality using a mixed methods secondary analysis.
    Methods: This mixed methods observational study performed a secondary analysis of a previously established dataset of publicly available TikTok videos related to mental health. A total of 207 videos addressing BPD or BPD-like traits were included. Quantitative analyses summarized video characteristics, engagement metrics, presenter backgrounds, content type, intent, and authenticity indicators using descriptive statistics. Qualitative data were analyzed using reflexive thematic analysis following Braun and Clarke's methodology to identify recurrent narratives related to BPD.
    Results: Among the 207 included videos, 185 (89.4%) were in English and 199 (96.1%) targeted the general public. Most creators (n=124, 59.9%) did not disclose professional credentials, whereas psychiatrists (n=18, 8.7%) and psychologists (n=16, 7.7%) represented the largest identified professional groups. Seven overarching themes were identified: understanding personality disorders, treatment and management, interpersonal relationships, myths and misconceptions, co-occurring disorders, personal experiences, and diagnosis and classification. Videos focused on personal experiences generated the highest audience engagement (mean weighted likes 2389, SD 6529; favorites 333, SD 707; shares 117, SD 287), whereas treatment-focused videos demonstrated substantially lower engagement (mean weighted likes 222, SD 360; favorites 54, SD 77; shares 19, SD 40). Opinion-based or mixed-content formats predominated across most themes. Reference-based information remained limited, ranging from 0% to 33.3% depending on thematic category, while misinformation was observed across multiple themes, particularly understanding personality disorders (20/81, 24.7%) and diagnosis and classification (14/48, 29.2%). Experiential and relational narratives were substantially more prominent and engaging than evidence-based psychoeducational content.
    Conclusions: TikTok discourse surrounding BPD is dominated by experiential narratives that foster engagement but frequently rely on opinion-based rather than evidence-based information. Although many videos may provide validation and reduce stigma, the limited use of references and the persistence of misinformation highlight the need for the greater involvement of mental health professionals in producing accessible, accurate, and engaging digital content. To our knowledge, these findings provide the first focused overview of BPD-related content on TikTok and may inform future strategies to improve the quality of online mental health information.
    Keywords:  TikTok; borderline personality disorder; digital psychiatry; health misinformation; mental health; personality disorders; psychoeducation; qualitative research; social media; thematic analysis
    DOI:  https://doi.org/10.2196/106246
  44. Front Digit Health. 2026 ;8 1880030
    Quality, reliability, and user engagement of hepatitis B-related short videos on TikTok, Bilibili, and Xiaohongshu: a cross-platform cross-sectional study.
    Lili Yu, Hongyu Wei, Yuanyuan Zhang.
       Background: CHB remains a major public health burden in China. TikTok, Bilibili, and Xiaohongshu have become key channels for health information, yet the quality, reliability, and transparency of such content remain uncertain.
    Methods: On November 12, 2025, the top 100 eligible HBV-related videos from each platform (N = 300) were assessed using the Global Quality Scale (GQS), modified DISCERN instrument (mDISCERN), and Journal of the American Medical Association benchmark criteria (JAMA). Two raters independently scored each video; discrepant ratings were resolved by consensus or senior adjudication, and final consensus scores were used for the main analyses. Inter-rater agreement was assessed using linearly weighted Cohen's κ. Group differences were evaluated using the Kruskal-Wallis H test followed, when significant, by Dunn's post hoc test with Holm adjustment. Spearman correlations and multivariable ordinal regression models were used to examine unadjusted and adjusted associations.
    Results: A total of 300 videos were included, with 100 from each platform. Inter-rater agreement was substantial for GQS (κ = 0.774, 95% CI 0.705-0.839), almost perfect for mDISCERN (κ = 0.928, 95% CI 0.888-0.962), and moderate for JAMA (κ = 0.503, 95% CI 0.410-0.590). mDISCERN and JAMA scores differed significantly across platforms, whereas GQS scores did not. Medical personnel had higher rank distributions than normal users for GQS, mDISCERN, and JAMA after Holm adjustment. In adjusted models, Xiaohongshu was associated with higher GQS scores than Bilibili, normal-user uploaders were associated with lower GQS scores than medical personnel, and longer video duration was associated with higher GQS scores. JAMA showed threshold-specific associations with platform, uploader type, and video duration. The adjusted cumulative mDISCERN model was not formally interpreted because of sparse-data instability; a bias-reduced adjacent-category model was reported as an exploratory sensitivity analysis. Engagement metrics showed weak and inconsistent correlations with quality scores.
    Conclusions: Among highly visible hepatitis B-related short videos retrieved under prespecified search conditions, video quality varied by assessment instrument, platform, and uploader type. Adjusted associations were instrument- or threshold-specific, and engagement metrics should not be treated as proxies for reliability or transparency.
    Keywords:  health information quality; hepatitis B; patient education; short video platforms; social media
    DOI:  https://doi.org/10.3389/fdgth.2026.1880030
  45. NPJ Prim Care Respir Med. 2026 Oct 22. pii: 69. [Epub ahead of print]36(1):
    Evaluating influenza-related short videos on Chinese platforms: information quality, user engagement, and public health implications.
    Xinlei Zhang, Chunyuan Zhou, Yuyang Hu, Junchao Yang.
      Influenza-related short videos have become an important source of health information on Chinese social media. In an ongoing "information disorder" environment, where digital misinformation is known to cause health harms, their information quality and influenza-specific public health implications on major short-video platforms remain insufficiently characterized. On October 7, 2025, new accounts were registered to search for short videos on TikTok and Bilibili using the keywords "" (influenza) and "" (influenza) respectively. Basic information was extracted from each video, and quality was assessed using JAMA, GQS, modified DISCERN, and PEMAT, which do not verify factual accuracy. Statistical analysis was performed using SPSS software. A total of 229 influenza-related short videos were included, primarily uploaded by doctors (n = 73, 31.9%) and non-professional individuals (n = 88, 38.4%). Disease knowledge constituted the main video content (n = 176, 76.9%), with treatment being the primary video theme(n = 85, 37.12%), and monologue the predominant video format (n = 84, 36.68%). Compared to Bilibili, TikTok videos featured shorter lengths, more recent upload dates, higher counts of likes, comments, and shares, and were posted by creators with larger fans counts (all P < 0.001). Bilibili scored significantly higher than TikTok on both JAMA (2.5 (2, 3)) and GQS (4 (4, 4)) scales (both P < 0.001); while TikTok scored higher on PEMAT comprehensibility (100 (82, 100)) and executability (67 (0, 67)) (both P < 0.001). Videos from physicians, news agencies, and non-profit organizations, and those focusing on treatment or diagnosis, generally showed higher quality. The four quality scales were positively correlated(all P < 0.05), while correlations between quality scores and engagement metrics were weak. In the context of ongoing information disorder, influenza-related short videos have become a major channel for public health information access, yet the overall quality of this content remains suboptimal, with deficiencies in key clinical and preventive points. Improving short-form science communication will require higher-quality content, stronger platform governance, and enhanced public media literacy.
    DOI:  https://doi.org/10.1038/s41533-026-00505-6
  46. J Health Commun. 2026 Oct 20. 1-10
    "A Minefield for the Novice": Australian Women's Experiences Seeking Health Information Online About Perimenopause.
    Nikki Boer, Natalie Jovanovski, Lauren Saling.
      During perimenopause, many women experience psychological and physiological symptoms that can seriously impact quality of life and future health. However, they report challenges accessing appropriate information and medical support, likely exacerbated by clinician knowledge gaps, stigma, and a persistent lack of research, consistent with a documented pattern of structural sexism in health systems. Consequently, women increasingly turn to alternative sources, particularly online, for support. Although scholarship exists on online health-information seeking and women's digital health engagement, limited research examines how women navigate online health information under clinical and structural invalidation. This study gathered responses from 58 perimenopausal women through an online qualitative survey and constructed three themes using reflexive thematic analysis: (a) the search for validation; (b) trust erosion and the support gap; and (c) self-efficacy, agency and kinship. Interpreted through an epistemic injustice lens, the findings indicated that participants used digital literacy strategies to reassert epistemic standing and compensate for structural lack of support, yet were left to navigate variable-quality health information. Therefore, health information providers and policymakers should acknowledge perimenopausal complexity, model knowledge-appraisal strategies rather than present information as resolved, and mandate training to address the support gap. Health professionals should also validate women's epistemic standing before clinically interpreting symptoms.
    Keywords:  Perimenopause; health communication; information-seeking behavior; online health information; women’s health
    DOI:  https://doi.org/10.1080/10810730.2026.2750443
  47. J Med Libr Assoc. 2026 Oct 01. 114(4): 410-418
    Pharmacist resource use across the United States: implications for teaching and practice.
    Rachel Whitney, Hilary Jasmin, Christopher Alan Giuliano, Audrey Banner Kostrzewa, Melissa Hunter, Christopher S Wisniewski, Emily F Gorman.
       Objective: To analyze practicing pharmacists' use of drug information (DI) resources in the United States.
    Methods: An electronic survey was distributed to practicing US pharmacists through professional organization email lists and social media. The survey questionnaire asked respondents which subscription-based DI resources are provided through their employer, whether they personally pay for additional resources, how often they use subscription-based resources, and other related questions.
    Results: The survey received 430 responses, 387 of which were able to be fully analyzed. The resources that respondents most frequently had access to through their employer were UpToDate (77.0%), Lexicomp (74.9%), and Micromedex (62.3%). Many respondents (40.8%) paid for subscriptions with personal funds, and 28.9% reported never experiencing access issues due to payment barriers. Respondents in community settings were more likely to have paid for resources with personal funds (57.1%, p<0.05) and have less access to resources (51.4%, p<0.05). In the open-text responses, the most desired resources were point-of-care tools (UpToDate) and DI databases (Lexicomp).
    Conclusions: US pharmacists experience a varied level of access to DI resources through their employer. Pharmacy educators should focus their instruction on resources that are the most accessible to practicing pharmacists. Many, especially community pharmacists, pay out of pocket for desired resources, presenting an opportunity to advocate for increased resources.
    Keywords:  Drug information; collection development; electronic resources; pharmacy education; pharmacy practice
    DOI:  https://doi.org/10.5195/jmla.2026.2424
  48. Bioinform Adv. 2026 ;6(1): vbag283
    Accelerating metadata annotation in collaborative research centers: a hybrid AI workflow for biomedical entities.
    Manuel Watter, Felix Engel, Aref Kalantari, Claudia Giuliani, Karin Schuller, Claus-Werner Franzke, Markus Sperandio, Harald Binder, Klaus Kaier.
       Background: Collaborative Research Centers rely on FAIR-compliant, richly structured metadata, yet manual annotation is a major bottleneck. We implemented a search-augmented large language model (LLM) workflow within a local research data management system to pre-annotate biomedical entities, using human-in-the-loop verification to ensure data quality.
    Methods: The pipeline uses Gemini 3 Pro in a two-step prompting strategy: (1) identify dataset deposits and stable identifiers in articles converted to Markdown; (2) extract structured fields from curated repository landing pages rendered via a headless browser. To handle a highly hierarchical metadata schema, we flattened the schema for prompting and remapped outputs to strict JSON with granular provenance tags. Authors received pre-filled metadata and could accept, edit, or delete entries (TP, FP, FN mapping). Performance metrics (precision, recall, F1) were estimated as proportions and synthesized via random-effects meta-analysis.
    Results: Among 51 screened articles, the LLM identified a repository deposit in 31; authors responded for 17 (55% of articles with an identified deposit; 33% of all screened articles), yielding 39 verified datasets. False positives were rare (mean 0.23) and false negatives low (mean 1.46). Precision was consistently high at 99.65% (95% CI 98.42%-100.00%). Recall showed more variability at 93.75% (95% CI 89.79%-96.96%), and the F1 score reached 96.17% (95% CI 93.78%-98.11%).
    Conclusions: The hybrid workflow achieved high author-verified precision with moderate recall, shifting effort from drafting to reviewing while maintaining schema compliance. As these metrics rely on author acceptance and response rates were modest, broader, multi-site validation is needed.
    DOI:  https://doi.org/10.1093/bioadv/vbag283
  49. J Med Libr Assoc. 2026 Oct 01. 114(4): 393-400
    Scholarly publishing literacy and perceptions of predatory publishing among educational partners at a chiropractic college: a case report.
    Jennifer Smith, Edward W Murphy, Stacie A Salsbury.
       Background: Chiropractors must find, appraise, and apply high quality, peer-reviewed information sources to make evidence-based clinical practice decisions. This case report describes perceptions of scholarly publishing practices, with a focus on predatory and deceptive publishing, among chiropractic educational partners at one chiropractic college.
    Case Presentation: An anonymous, online, cross-sectional survey was administered to a convenience sample of 2,476 faculty, staff, and students. The 35-item questionnaire elicited publishing experience, critical appraisal skills, knowledge of scholarly and predatory publishing, and perceptions of chiropractic journals and information resources. Respondents (n=204, 111 students, 36 faculty, 16 administrators, and 41 staff) provided usable survey responses, though skipped items were common. Twenty-six percent had published one or more peer-reviewed publications. Most self-rated themselves as novice or intermediate in their ability to perform literature searches (87%) or critically appraise research articles (82%). Forty-six percent to 69% had not read or were unsure of the quality of scientific evidence for 11 chiropractic-related journals. Fifty-seven percent were unsure or unfamiliar with predatory journals and 51% lacked confidence in recognizing these journals. Predatory practices identified included insufficient or fast peer review, editorial board and impact factor issues, publication fees, plagiarized articles, and unsolicited request for paper submissions. Sixty-one percent requested training or resources on identifying quality academic journals for publication or research.
    Conclusion: Chiropractic educational partners were unfamiliar with most chiropractic journals and had little familiarity with predatory publishing. Targeted training on identifying quality academic publications, conducting literature reviews and critically appraising research articles is recommended.
    Keywords:  Chiropractic; Education, Professional; Information Literacy; Libraries, Medical; Library Surveys; Predatory Journals as Topic; Scholarly Communication
    DOI:  https://doi.org/10.5195/jmla.2026.2226
This page is being maintained by Thomas Krichel. It was last updated on 2026‒10‒25 at 5:06.