bims-librar Biomed News
on Biomedical librarianship
Issue of 2026–11–15
thirty-six papers selected by
Thomas Krichel, Open Library Society



  1. Med Ref Serv Q. 2026 Nov 10. 1-7
      By situating library resources and services within the levels of the 6S pyramid, librarians can demonstrate how their work supports efficient access to the most appropriate evidence available. This commentary paper explores how aligning library instruction, resource organization, and service delivery with evidence-based practice frameworks can help bridge the gap between clinicians' information needs and the evidence ecosystem that supports patient care.
    Keywords:  Evidence pyramid; hospital library; reference services
    DOI:  https://doi.org/10.1080/02763869.2026.2758118
  2. Nucleic Acids Res. 2026 Nov 14. pii: gkag1151. [Epub ahead of print]
      The Protein Data Bank in Europe (PDBe; pdbe.org) provides open access to experimentally determined biomacromolecular structures and enriches these data with biological and functional context. Here, we describe recent developments that expand the information connected to PDB structures and introduce new ways for researchers and computational tools to access it. Literature integration has been extended to additional open-access publications, substantially increasing the availability of residue-level annotations mined from the scientific literature. Cross-links to molecular dynamics simulations in MDposit have been established, and simulation-derived residue-level annotations, including relative solvent accessibility and membrane-related properties, have been integrated into PDBe-KB (pdbekb.org). PDBe entry pages now expose additional experimental and sequence information, links to related resources, and extended PDB identifiers. Protein channel, tunnel, and pore analyses generated using MOLE have also been incorporated into our weekly release pipeline and made available for download. Improved API documentation facilitates the discovery of data fields and search parameters, while three Model Context Protocol servers facilitate agentic access to the PDBe APIs, Search, and the PDBe-KB knowledge graph. Together, these developments make structural data and its wider biological context easier to discover, interpret, and reuse through both conventional and agentic workflows.
    DOI:  https://doi.org/10.1093/nar/gkag1151
  3. Nucleic Acids Res. 2026 Nov 14. pii: gkag1159. [Epub ahead of print]
      The National Center for Biotechnology Information (NCBI) provides three principal resources containing DNA, RNA, and protein sequence data: the Sequence Read Archive (SRA), GenBank®, and the Reference Sequence (RefSeq™) collection. All resources are free to the public. SRA and GenBank are part of the International Nucleotide Sequence Database Collaboration and share data daily. After summarizing the current state of these resources, we review significant updates in the past two years, including these to the Submission Portal, the primary interface for submitting data to SRA and GenBank. We also describe access points for these data including NCBI Datasets' website and programmatic interface, and present improvements in NCBI's main sequence analysis and visualization tools, such as annotation pipelines, BLAST, and the Multiple Comparative Genome Viewer.
    DOI:  https://doi.org/10.1093/nar/gkag1159
  4. J Biomed Semantics. 2026 Aug 12. pii: 24. [Epub ahead of print]17(1):
       BACKGROUND: The rapid expansion of biomedical literature has made comprehensive manual synthesis increasingly difficult, creating a pressing need for AI systems capable of reasoning over verified evidence rather than merely retrieving it. Existing retrieval-augmented generation (RAG) methods often struggle with complex biomedical queries that require accurate citation, iterative reasoning, and multi-step evidence synthesis. To address this challenge, we developed Queryome, a deep research system composed of specialized large language model (LLM) agents that dynamically adapt their orchestration to the complexity of each query. Queryome operates on a hybrid semantic-lexical retrieval engine covering 28.3 million PubMed abstracts and performs iterative, evidence-grounded synthesis through a hierarchical multi-agent architecture.
    RESULTS: On the MIRAGE benchmark, Queryome achieved 89.93% accuracy, outperforming the best prior system by up to 9.96% points. On the biomedical subset of Humanity's Last Exam (HLE), it improved reasoning accuracy from 14.6% to 24.9% relative to the base reasoning model. In a review article construction task, Queryome achieved the highest composite score among deep research systems from OpenAI, Google, Perplexity, and Scite.ai, demonstrating strong capabilities in literature retrieval and synthesis.
    CONCLUSIONS: Queryome demonstrates that hierarchical, dynamically orchestrated multi-agent systems can substantially enhance evidence-based reasoning and synthesis over biomedical literature. Its consistent performance gains across diverse benchmarks suggest that agentic RAG architectures offer a promising path toward reliable, citation-grounded biomedical knowledge synthesis at scale.
    Keywords:  Agentic AI; Agentic RAG; Deep research; Large language models; Literature mining
    DOI:  https://doi.org/10.1186/s13326-026-00366-5
  5. Bioinformatics. 2026 Nov 12. pii: btag814. [Epub ahead of print]
       SUMMARY: Public omics databases like the Gene Expression Omnibus and the Sequence Read Archive offer substantial opportunities for data reuse to address novel biomedical questions. However, it is still difficult to find samples and studies of interest since they are described by free-text metadata and lack standardized annotations. To address this issue, multiple research groups have undertaken curation efforts to add standardized annotations to large collections of these data, but these annotations are fragmented across online resources and are stored in different formats subject to varying standardization criteria, hindering the integration of annotations across sources. We developed MetaHQ to harmonize and distribute standardized metadata for public omics samples. MetaHQ comprises a database with annotations for >567k samples and >27k studies in GEO from 13 sources and a user-friendly command-line interface (CLI) to query the database and retrieve annotations.
    AVAILABILITY AND IMPLEMENTATION: The MetaHQ CLI is deployed as a Python Package on PyPI at https://pypi.org/project/metahq-cli that accesses the MetaHQ database available at https://doi.org/10.5281/zenodo.17663086. Project source code and documentation are available at https://github.com/krishnanlab/meta-hq.
    SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
    DOI:  https://doi.org/10.1093/bioinformatics/btag814
  6. Nucleic Acids Res. 2026 Nov 14. pii: gkag1164. [Epub ahead of print]
      PubChem (https://pubchem.ncbi.nlm.nih.gov) is a public chemical information resource that serves scientific communities across various domains, as well as the general public. Over the past 2 years, substantial changes have been made to PubChem. The addition of over 100 new data sources has broadened its data coverage. A new data collection, RefChem, containing chemical records compiled by the PubChem team from authoritative and curated sources, was introduced to improve annotation mapping and disambiguation and to provide better access to information on substances with non-discrete structures. The Quick Search supports general exploration across multiple data collections, while the Advanced Search provides more structured, fielded search options for individual collections. The new Search History page enables users to formulate complex queries using Boolean operators. Major updates were also made to the Classification Browser and Co-occurrence Panels. Updates to PubChemRDF improved its search performance and interoperability. Practical guidance was provided for its efficient use in cloud environments. This article provides an overview of the latest updates to PubChem.
    DOI:  https://doi.org/10.1093/nar/gkag1164
  7. J Obstet Gynaecol. 2026 Dec;46(1): 2758484
       BACKGROUND: Freely accessible large language models provide health information, but their communication characteristics require evaluation. We compared information quality, transparency, readability and perceived empathy for questions about polyendocrine metabolic ovarian syndrome (PMOS; formerly PCOS).
    METHODS: Thirty-three clinician-formulated English questions using PCOS terminology were submitted once to each of three free-access configurations during 20-25 July 2026. Two clinicians rated DISCERN, EQIP, JAMA benchmarks, global quality and perceived empathy. Six readability indices were calculated. Matched comparisons used Friedman tests, paired signed-rank permutation comparisons and Holm correction.
    RESULTS: Across 99 archived responses, the DeepSeek-V4 configuration received higher final ratings than GPT-5.5 and Gemini 3.5 on all five clinician-rated measures (Holm-adjusted pairwise P < 0.05). Median DISCERN scores were 67.0, 62.0 and 61.0, respectively. DeepSeek-V4 had higher median ARI, GFI, FKGL and SMOG values, whereas GPT-5.5 had the highest median Coleman-Liau score. FRES showed no overall difference (adjusted P = 0.100). Initial empathy agreement was low (ICC 0.05-0.43).
    CONCLUSION: Ratings differed for the specific configurations and single responses examined in this treatment-heavy question set. The findings neither establish stable model rankings nor demonstrate clinical correctness, safety or effectiveness for patients.
    Keywords:  Polyendocrine metabolic ovarian syndrome; health communication; large language models; patient education; reproductive health
    DOI:  https://doi.org/10.1080/01443615.2026.2758484
  8. JMIR Med Inform. 2026 Nov 09. 14 e104836
       BACKGROUND: Large language models (LLMs) are increasingly used for health information, but chronic obstructive pulmonary disease (COPD)-related questions often involve safety-sensitive decisions, including treatment escalation, oxygen therapy, and exacerbation management. Because generative outputs may vary across repeated prompts, response stability is also relevant to comparative evaluation.
    OBJECTIVE: This study aims to compare 6 public-facing LLM chatbots in answering COPD-related public health questions across safety, accuracy, empathy, information quality, source transparency, readability, and 2-round response stability.
    METHODS: We developed a 55-question English COPD question bank from an initial 108-question pool using web-based retrieval, Google Trends, PubMed-informed topic review, patient education resources, and expert screening. Six public-facing chatbot products were evaluated in a primary first round and a post hoc independent second round, added after first-round data collection, using the same 55 questions. Visible model/mode labels are provided in the supplementary materials. Five blinded raters assessed manually scored outcomes against a Global Initiative for Chronic Obstructive Lung Disease (GOLD) 2026-centered reference standard, and 6 readability indices were calculated. The primary cross-model analysis used the first round; the second round was analyzed as a matched retest sensitivity analysis. Stability was assessed using exact agreement and Cohen κ for safety, quadratic-weighted κ for ordinal ratings, and absolute-agreement single-measure intraclass correlation coefficients (ICCs) for continuous scores and readability indices; 95% CIs used question-level bootstrap resampling.
    RESULTS: The first round yielded 330 responses. Unsafe responses occurred in every model: ChatGPT, Gemini, Copilot, and DeepSeek each generated 4 unsafe responses (7.3%); Qwen generated 5 (9.1%); and Doubao generated 6 (10.9%). The overall safety comparison was not statistically significant (Cochran Q=1.909, df=5; P=.86); this finding should not be interpreted as evidence of equivalent safety. Accuracy, empathy, information-quality measures, source-transparency scores, and readability indices differed significantly across models. Across 330 matched question-model pairs, safety classifications agreed in 97.3% of pairs (κ=0.821, 95% CI 0.543-0.979); accuracy and empathy showed weighted κ values of 0.872 and 0.856, respectively; DISCERN (an instrument for judging the quality of written consumer health information on treatment choices) and adapted Ensuring Quality Information for Patients (EQIP) showed ICCs of 0.954 and 0.925. Most readability indices also showed high overall 2-round stability, although model-specific variability remained. Nine safety classifications changed across the 2 rounds (5 from safe to unsafe and 4 from unsafe to safe).
    CONCLUSIONS: In this standardized single-turn COPD question set, model performance differed across several evaluated outcomes, whereas the first-round safety test neither established a between-model difference nor demonstrated equivalence. All 6 models produced potentially unsafe advice, and 9 safety classifications changed between rounds conducted several months apart despite high pooled agreement. These findings support cautious use of LLMs for supplementary COPD education, with clinician oversight for individualized treatment decisions and ongoing reassessment as public-facing systems change.
    Keywords:  CHART; COPD; Chatbot Assessment Reporting Tool; LLM; chatbot; chronic obstructive pulmonary disease; health information quality; large language model; patient education; readability; safety
    DOI:  https://doi.org/10.2196/104836
  9. BMC Pediatr. 2026 Aug 12. pii: 1141. [Epub ahead of print]26(1):
       BACKGROUND: Access to accurate and understandable information for parents in pediatric dentistry is critically important. This study aimed to evaluate the information reliability and readability of nine different artificial intelligence (AI) chatbots, referencing the American Academy of Pediatric Dentistry (AAPD) FAQs.
    METHODS: Responses to AAPD FAQs from AI chatbots (ChatGPT-4o, ChatGPT-4.1, ChatGPT-5, Gemini 2.5 Pro, Claude 4, Perplexity, Grok 4, DeepSeek V3, LLaMA) were independently evaluated by two experts using the DISCERN questionnaire for information reliability and the Flesch Reading Ease Score (FRES) and Flesch-Kincaid Grade Level (FKGL) metrics for readability. Given the repeated-measures (within-question) design, the Friedman test with post-hoc Wilcoxon signed-rank comparisons (Bonferroni-corrected) was used for statistical analyses.
    RESULTS: Statistically significant differences (p < 0.05) were found among AI models in both information reliability (DISCERN) and readability (FRES, FKGL) scores. Perplexity provided the highest information reliability (DISCERN: 74.93 ± 4.79), while Grok 4 yielded the highest readability (FRES: 78.36 ± 11.04); however, Grok 4 also had the lowest reliability score (DISCERN: 46.40 ± 8.48). Claude 4 and DeepSeek V3 demonstrated balanced and above-average performance across all metrics. No AI model achieved the best performance simultaneously across all metrics.
    CONCLUSIONS: AI chatbots hold clear potential for informing parents in pediatric dentistry, but reliability and readability differed across models, with none performing best on both. An apparent readability-reliability trade-off in some models did not reach statistical significance and should be read as descriptive rather than causal. Clinicians should consider each model's strengths and weaknesses, verify information from independent sources, and position AI as a supportive tool. Models exhibiting balanced performance in the present evaluation, such as Claude 4 and DeepSeek V3, may represent promising options for clinical applications, although model performance may change with future updates.
    Keywords:  Artificial Intelligence; DISCERN; Flesch Reading Ease; Flesch-Kincaid; Frequently Asked Questions; Health Literacy; Information Reliability; Parent Education; Pediatric Dentistry; Readability
    DOI:  https://doi.org/10.1186/s12887-026-07533-8
  10. Healthcare (Basel). 2026 Nov 03. pii: 3726. [Epub ahead of print]14(21):
       BACKGROUND/OBJECTIVES: Large language models (LLMs) are increasingly used to obtain health information about musculoskeletal conditions. However, their performance across different types of clinical questions remains insufficiently understood. This study evaluated the accuracy, quality, information reliability, and readability of responses generated by three widely used LLMs (ChatGPT 5.2, Gemini 3, and DeepSeek V3.2) to questions about first-time traumatic anterior shoulder dislocation, with separate analyses of medical and rehabilitation questions.
    METHODS: Sixty-five clinically relevant questions were categorized into Medical and Safety (n = 28) and Rehabilitation (n = 37) domains. Responses were independently evaluated by domain experts using a 5-point accuracy scale, the Global Quality Scale (GQS), modified DISCERN (mDISCERN), and Flesch Reading Ease (FRE). Model differences were analyzed using Friedman tests with post hoc pairwise comparisons, and inter-rater agreement was assessed using intraclass correlation coefficients.
    RESULTS: In the Medical and Safety Domain, the models differed significantly in accuracy (p < 0.001), quality (p < 0.001), information reliability (p = 0.004), and readability (p = 0.011). ChatGPT 5.2 and Gemini 3 had higher accuracy and GQS scores than DeepSeek V3.2, while DeepSeek V3.2 showed significantly higher mDISCERN scores than ChatGPT 5.2. Gemini 3 demonstrated significantly higher FRE scores than DeepSeek V3.2. In the Rehabilitation Domain, no significant differences were found in accuracy, quality, or information reliability (all p > 0.05), whereas readability differed significantly (p < 0.001), with ChatGPT 5.2 and Gemini 3 showing higher FRE scores than DeepSeek V3.2. Median accuracy and GQS scores were 5 for all three models in the Rehabilitation Domain, suggesting a potential ceiling effect. Readability was relatively difficult across the models, with no statistically significant correlations between FRE scores and accuracy, GQS, or mDISCERN scores. Inter-rater agreement ranged from moderate to excellent.
    CONCLUSIONS: LLMs provided generally accurate and high-quality information about first-time traumatic anterior shoulder dislocation. In separate domain-specific analyses, significant differences among models were observed in accuracy, quality, information reliability, and readability for medical and safety questions. For rehabilitation questions, statistically significant differences were detected only in readability. These findings do not establish a statistically significant difference in model performance between the two clinical domains. LLM-generated information should be interpreted cautiously, particularly when addressing medical decisions and patient safety.
    Keywords:  artificial intelligence; large language models; readability; rehabilitation; shoulder dislocation; sports medicine
    DOI:  https://doi.org/10.3390/healthcare14213726
  11. Front Public Health. 2026 ;14 1985180
       Background: Large language models (LLMs) are capable of providing health information and consultation to patients with osteonecrosis of the femoral head (ONFH). However, owing to imbalances in the linguistic distribution of their training corpora, the quality, readability, and safety profiles of outputs generated by the same model may vary across language settings.
    Objective: This study aimed to compare bilingual responses to ONFH patient queries generated by six state-of-the-art flagship LLMs, comprising three English-oriented models (ChatGPT-5.6-Sol, Gemini-3.1-Pro, and Claude-Fable-5) and three Chinese-oriented models (DeepSeek-V4-Pro, Qwen3.7-Max, and Doubao-Seed-2.1-Pro), and to multidimensionally evaluate their cross-linguistic capabilities for generating medical information in terms of information quality, actionability, clinical accuracy, safety risk, hallucination, and readability.
    Methods: A question set comprising 36 semantically distinct patient queries regarding ONFH was developed, yielding 72 semantically matched English-Chinese paired prompts. Six state-of-the-art flagship LLMs were evaluated. Information quality was assessed using the Global Quality Score (GQS), DISCERN, the Quality Assessment of Medical Artificial Intelligence (QAMAI), and the 20-item Ensuring Quality Information for Patients (EQIP-20). Actionability was assessed using the Patient Education Materials Assessment Tool for Printable Materials (PEMAT-P), while clinical accuracy was evaluated using a 5-point Likert scale. Safety risk was assessed using a graded classification framework, and prespecified criteria were established for identifying hallucinations. Readability was evaluated using language-specific readability metrics for English and Chinese responses. Statistical analyses included the Friedman test, paired Wilcoxon signed-rank test, and Spearman correlation analysis.
    Results: A total of 432 responses generated by LLMs were included in this study, and good-to-excellent inter-rater reliability was observed for all evaluation indicators. Significant between-model differences in actionability and clinical accuracy were identified among the six models under both English and Chinese response conditions (all p < 0.05). Furthermore, clinical accuracy exhibited high cross-linguistic stability. ChatGPT-5.6-Sol showed high actionability and a relatively low frequency of Level 1 minor safety-related omissions under the English-language condition, whereas switching to Chinese was accompanied by reduced actionability and an increased frequency of Level 1 omissions. All observed nonzero safety events were Level 1 minor omissions; no Level 2 or Level 3 unsafe responses were identified in either language condition. Responses generated by all six models in both languages failed to reach the prespecified readability threshold for patient education.
    Conclusion: The six LLMs showed significant inter-model differences in the actionability and clinical accuracy of responses to ONFH-related patient health queries in both English and Chinese settings, while clinical accuracy demonstrated favorable cross-linguistic stability. Notably, none of the six-model outputs met the applicable readability threshold for patient-targeted health education under either language condition. In clinical practice, model selection, prompting-language effects and hallucination-related risks should be comprehensively considered when deploying LLMs to deliver health information for patients with ONFH. At the present stage, LLM-generated content intended for direct patient-facing education requires cautious evaluation and human oversight.
    Keywords:  artificial intelligence; cross-linguistic analysis; large language models; osteonecrosis of the femoral head; patient education
    DOI:  https://doi.org/10.3389/fpubh.2026.1985180
  12. JMIR Form Res. 2026 Nov 10. 10 e93849
       Background: Hard flaccid syndrome (HFS) is a rare and poorly described andrological condition characterized by a persistent semirigid penis in the flaccid state and frequently accompanied by erectile dysfunction, sensory disturbances, urinary symptoms, and psychosocial distress. Due to the absence of standardized diagnostic and therapeutic guidelines, patients increasingly seek health information from AI-based tools such as ChatGPT. However, the reliability of such tools in rare diseases remains uncertain.
    Objective: This study aimed to perform an exploratory evaluation of the accuracy, readability, and short-term reproducibility of ChatGPT responses to patient-generated questions regarding HFS.
    Methods: Sixteen clinically relevant questions spanning 4 domains (epidemiology, treatment, diagnostics, and symptoms and case presentation) were developed using authentic patient-derived queries and previously published AI evaluations in rare diseases. Questions were submitted to the ChatGPT web interface (OpenAI) in separate chat sessions and repeated after 48 hours to assess short-term reproducibility. Three urologists with expertise in male sexual health independently evaluated response accuracy using a validated 4-point scale. Readability was assessed using the Gunning Fog Index, whereas interrater reliability was evaluated using intraclass correlation coefficients (ICCs).
    Results: Overall readability remained stable between sessions, with mean Gunning Fog Index scores of 15.19 (SD 1.68) at baseline and 14.59 (SD 1.66) after 48 hours (P=.33), corresponding to an upper high school to early college reading level. No significant differences in readability were observed across domains at either time point (T1 [November 27, 2025]: P=.76; T2 [December 1, 2025]: P=.68). Accuracy differed across clinical domains. At T1, a total of 43.8% (7/16) of the responses were classified as correct and comprehensive, 50% (8/16) were classified as correct but inadequate, and 6.3% (1/16) were classified as having mixed correct and incorrect information; no responses were classified as completely incorrect. At T2, performance improved, with 62.5% (10/16) of responses classified as correct and comprehensive and 37.5% (6/16) classified as correct but inadequate. Interrater reliability demonstrated fair agreement at the individual reviewer level and moderate agreement when ratings were averaged (T1 single-measure ICC=0.36, 95% CI 0.06-0.67; T1 average-measure ICC=0.63, 95% CI 0.17-0.86; T2 single-measure ICC=0.33, 95% CI 0.04-0.64; T2 average-measure ICC=0.59, 95% CI 0.10-0.84). Short-term reproducibility was substantial, with 87.5% (14/16) of responses receiving identical final accuracy classifications between sessions.
    Conclusions: ChatGPT demonstrated stable readability and high short-term consistency when addressing HFS-related questions. However, accuracy remained variable across clinical domains, with responses occasionally extending beyond the available evidence base in this poorly characterized condition. These findings suggest that AI tools may serve as accessible informational resources but should complement rather than replace expert clinical evaluation and evidence-based guidance, particularly in rare diseases for which clinical consensus remains limited.
    Keywords:  AI; ChatGPT; artificial intelligence; hard flaccid syndrome; medical information accuracy; rare andrological disorders
    DOI:  https://doi.org/10.2196/93849
  13. Front Med (Lausanne). 2026 ;13 2007153
       Background: Large language models (LLMs) may support patient education, but their suitability for diabetic kidney disease (DKD) remains uncertain. We compared six publicly accessible LLM chatbots-ChatGPT, Gemini, DeepSeek, Doubao, Kimi, and Qwen-across safety, accuracy, empathy, information quality, and readability, with safety as the primary outcome.
    Methods: In this CHART-guided cross-sectional study, 42 standardized English DKD questions were submitted once to each chatbot in independent single-turn sessions, generating 252 responses. Five source-masked clinical raters assessed safety, accuracy, empathy, DISCERN, EQIP, JAMA benchmarks, and the Global Quality Scale; readability was assessed automatically. Agreement was evaluated using Fleiss'κ and ICC (2,1)/ICC (2,5). Model differences were tested using Cochran's Q or Friedman tests with Kendall's W and Benjamini-Hochberg correction. Potentially unsafe responses underwent qualitative review.
    Results: Safety showed substantial agreement (κ = 0.854); ICC (2,1) ranged from 0.552 to 0.855 and ICC (2,5) from 0.860 to 0.967. Safe-response rates ranged from 83.3% to 92.9% without significant between-model differences (P = 0.726). Accuracy and empathy differed significantly (both P < 0.001), with ChatGPT highest for accuracy and Gemini highest for clinician-rated empathy. Information quality also differed significantly, with ChatGPT performing best overall and Doubao lowest. All six readability indices differed across models (all P < 0.001; Kendall's W = 0.139-0.255). DeepSeek had the lowest mean scores on all five grade-level indices and the highest mean FRES, but mean grade-level estimates exceeded the sixth-grade reference level for every model. Thirty of 252 responses (11.9%) were potentially unsafe, most often because of incomplete medication-related information or insufficiently individualized fluid advice.
    Conclusions: Publicly accessible LLMs can provide plausible DKD information but show variable performance and safety risks. They may support, but should not replace, professional patient education; guideline anchoring, source verification, improved readability, and clinical oversight remain essential.
    Keywords:  artificial intelligence; diabetic kidney disease; large language models; patient education; safety
    DOI:  https://doi.org/10.3389/fmed.2026.2007153
  14. Front Public Health. 2026 ;14 1961492
       Objective: Clarify the effects of large-language-model types and health-education content categories on generated-text quality (patient-education suitability, overall quality) and readability, so as to provide empirical evidence for standardized AI-assisted health-communication applications.
    Methods: This cross-sectional study was conducted between July 10 and July 15, 2026. Twenty herpes zoster-related questions were developed by a dermatologist and a pharmacist and categorized into five themes. The questions were then entered into five mainstream large language models (GPT-5, Doubao, DeepSeek, Qianwen, and Gemini), generating 100 textual responses. Response quality was assessed using the Chinese version of the Patient Education Materials Assessment Tool for Printable Materials (C-PEMAT-P) and Global Quality Score (GQS). Readability was evaluated via seven indices including the Automated Readability Index (ARI) and Flesch Reading Ease Score (FRES). Readability outcomes were compared using one-way repeated-measures analysis of variance with Greenhouse-Geisser correction. C-PEMAT-P raw totals and ordinal GQS scores were compared using Friedman tests. Post-hoc comparisons were adjusted using the Holm method. Category-level comparisons were performed using linear mixed-effects models with likelihood-ratio χ2 tests for continuous outcomes and an ordinal generalized estimating equation with a Wald χ2 test for GQS.
    Results: Significant overall differences among models were observed for all seven readability indices and both assessment scores (all p < 0.001). Median C-PEMAT-P raw total scores were 12.50, 11.00, 10.00, 9.00 and 9.00 for GPT-5, Doubao, DeepSeek, Qianwen and Gemini, respectively; corresponding median GQS scores were 5.00, 4.00, 4.00, 3.00, and 3.00. After Holm correction, GPT-5 achieved higher scores than each of the other models on both assessment measures. However, no significant difference in C-PEMAT-P raw totals was found between DeepSeek and either Qianwen or Gemini, or between Doubao and DeepSeek, after adjustment (all adjusted p = 0.062), and Qianwen and Gemini did not differ significantly (adjusted p = 0.766). The overall effect sizes were Kendall's W = 0.611 for C-PEMAT-P and W = 0.779 for GQS. Exploratory pooled Spearman correlation analysis revealed a positive association between the two assessment scores (ρ = 0.853).
    Conclusion: Model type is a key determinant of text quality. Readability was independently influenced by both model type and content category, and the two are relatively independent. In clinical applications, GPT-5, DeepSeek, and Doubao are preferred. Readability can be optimized by integrating topic characteristics and patients' health literacy to enhance communication accuracy. Limitations exist in this study regarding sample coverage and evaluation dimensions (the absence of subjective experience data); larger-scale verification should be conducted in future research.
    Keywords:  global quality score; herpes zoster; large language models; patient education; readability
    DOI:  https://doi.org/10.3389/fpubh.2026.1961492
  15. Diagn Interv Radiol. 2026 Nov 12.
       PURPOSE: Although professional societies provide standardized resources, the reliability of online health content remains inconsistent, particularly as large language models (LLMs) become a growing source of patient information. Therefore, the primary objective of this study was to determine whether patient-focused prompt optimization improves the readability of Generative Pre-trained Transformer-5 (GPT-5) Thinking responses compared with standard prompting while evaluating potential changes in expert-rated factual accuracy using educational materials published by the European Society of Breast Imaging (EUSOBI).
    METHODS: All GPT-5 Thinking outputs were prospectively generated in January 2026. The dataset included frequently asked questions (FAQs) and special information notes (SINs) derived from EUSOBI articles, each submitted under two distinct prompts (standard-Prompt 1 vs. patient-focused-Prompt 2). Readability was quantified using four indices with an averaged composite "reading grade level." Accuracy was scored by two radiologists using a 5-point Likert scale.
    RESULTS: A total of 48 questions (15 FAQs and 33 SIN-derived questions) were each answered under two prompt conditions, yielding 96 responses for analysis. Prompt 2 produced significantly lower readability levels than Prompt 1 across all indices (P < 0.001). These improvements occurred without significant differences in word count (P = 0.12) or sentence count (P = 0.63). The mean accuracy scores remained high and comparable between prompts for all questions (4.54 ± 0.63 vs. 4.52 ± 0.59, P = 0.75). The inter-rater agreement was almost perfect for both the FAQ and SIN-derived question sets (κ = 0.82 and 0.81, respectively).
    CONCLUSION: Patient-focused prompting substantially reduced the linguistic complexity and reading grade level of GPT-5 Thinking responses without a statistically significant reduction in expert-rated accuracy, demonstrating its ability to produce more accessible breast imaging content, although the resulting grade level remained above widely recommended patient-education thresholds.
    CLINICAL SIGNIFICANCE: A simple, low-cost prompt modification can substantially improve the readability of LLM-generated patient education content in breast imaging without a measurable reduction in expert-rated accuracy, although further work is needed to confirm this translates into improved patient comprehension.
    Keywords:  ChatGPT; Women’s health; artificial intelligence; breast imaging; health literacy; large language model; patient education; readability
    DOI:  https://doi.org/10.4274/dir.2026.264309
  16. Front Public Health. 2026 ;14 1990753
       Background: Large language models (LLMs) are increasingly used by patients and caregivers to find health information. However, their performance in providing Simplified Chinese education on pediatric pectus excavatum (PE) remains uncertain. This study comprehensively evaluated the responses generated by five mainstream LLMs.
    Methods: Thirty PE-related frequently asked questions (FAQs) spanning five clinical domains were collected and submitted in Simplified Chinese to ChatGPT, Claude, DeepSeek, Qwen, and Kimi. Responses were evaluated by pediatric thoracic surgeons from three institutions for accuracy, completeness, and relevance, using a 6-point Likert scale. Readability was assessed using AlphaReadabilityChinese (ARC) 2.0 and the Chinese Readability Index Explorer (CRIE) 3.0. Understandability, actionability, and overall quality were evaluated with the Chinese Patient Education Materials Assessment Tool for Printable Materials (C-PEMAT-P) and the Global Quality Score (GQS). Friedman tests compared the models, followed by Holm-adjusted Wilcoxon signed-rank tests.
    Results: All models achieved mean accuracy, completeness, and relevance scores above 4.9 on the 6-point scale, but the leading model differed by dimension. Claude had the highest mean accuracy (5.39 ± 0.28), completeness (5.22 ± 0.22), and GQS (4.80 ± 0.41), while DeepSeek had the highest relevance (5.42 ± 0.27). Qwen was the least readable: its median CRIE grade was 7.0, compared with 5.5 to 6.0 for the other models. Only 23.3% of its responses were at Grades 1-6, compared with 66.7-86.7% for the others. Overall C-PEMAT-P scores also differed (p = 0.003), with Kimi showing the highest numerical mean total (12.17 ± 0.75). Qwen combined the highest actionability (2.57 ± 0.50) with the lowest understandability (9.47 ± 0.57) and GQS (3.10 ± 0.71).
    Conclusion: Leading LLMs generally provided accurate, complete, and relevant Simplified Chinese answers to common pediatric PE questions, but no model performed best on every dimensions. LLM outputs should therefore be regarded as a supervised educational aid that require expert verification and adaptation to the health literacy of the intended audience, rather than as substitutes for specialist counseling.
    Keywords:  children; large language models; patient education; pectus excavatum; quality; readability
    DOI:  https://doi.org/10.3389/fpubh.2026.1990753
  17. Int J Gynaecol Obstet. 2026 Nov 09.
       BACKGROUND: Patients increasingly use large language model (LLM)-based chatbots to seek health information. Adenomyosis is a common gynecological condition associated with heavy menstrual bleeding, dysmenorrhea, pelvic pain, infertility, and pregnancy-related concerns. The quality and safety of chatbot-generated patient information is clinically relevant.
    OBJECTIVE: Our objective was to compare the safety, accuracy, empathy, reliability, information quality, and readability of LLM responses to patient health questions about adenomyosis.
    METHODS: This comparative cross-sectional evaluation was designed and reported according to the CHART checklist for chatbot health advice studies. Standardized chatbot queries were posed from June 13 to June 15, 2026, in Wuhan, China. A 40-item adenomyosis patient question set was submitted to four LLM-based chatbots: ChatGPT, Gemini, DeepSeek, and Doubao. Five trained gynecology experts evaluated the safety, accuracy, empathy, DISCERN score, EQIP (Ensuring Quality Information for Patients) score, Journal of the American Medical Association (JAMA) benchmark criteria, and Global Quality Scale score using prespecified criteria; readability was calculated algorithmically using automated text-processing procedures. Safety was coded as 0 for safe and 1 for potentially unsafe. Inter-rater agreement was assessed using Fleiss' kappa for safety and intraclass correlation coefficient (2,1) for the remaining expert-rated metrics. Between-model comparisons used paired nonparametric methods, with Kendall's W and 95% confidence intervals reported as the omnibus effect-size measure for Friedman tests.
    RESULTS: A total of 160 model-question responses were evaluated. Inter-rater agreement was substantial to excellent across metrics, with Fleiss' kappa of 0.892 for safety and intraclass correlation coefficient values ranging from 0.768 to 0.886 for the remaining metrics. Most responses were classified as safe; observed safe-response proportions differed descriptively across models, and no pairwise safety comparison remained statistically significant after Benjamini-Hochberg adjustment. Accuracy and empathy differed significantly among models (both P < 0.001): ChatGPT achieved the highest median accuracy score. DISCERN, EQIP, and JAMA scores differed significantly among models, whereas Global Quality Scale scores did not. All readability indices differed significantly among models: Doubao had the highest grade-level readability scores and the lowest Flesch Reading Ease score, whereas Gemini had comparatively easier readability.
    CONCLUSIONS: LLMs produced generally safe and moderately high-quality responses to adenomyosis-related patient questions, but performance varied by model and by evaluation domain. These findings support the need for transparent, multidimensional assessment of LLM-generated health advice before patient-facing use.
    Keywords:  accuracy; adenomyosis; chatbot health advice; empathy; large language models; patient education; readability
    DOI:  https://doi.org/10.1002/ijgo.71446
  18. Cureus. 2026 Oct;18(10): e117785
       BACKGROUND: Social media platforms increasingly influence patient perceptions regarding reconstructive procedures. This study evaluated the characteristics, quality, engagement patterns, and user sentiment associated with breast reduction-related content on TikTok, while assessing the performance of artificial intelligence (AI) in content and sentiment analysis.
    METHODS: A cross-sectional analysis of 140 TikTok videos on breast reduction was conducted. Engagement metrics, including likes, comments, shares, saves, and a weighted engagement index, were collected. The modified DISCERN (mDISCERN) and Global Quality Scale (GQS) were used to appraise content quality by AI and two human reviewers. Sentiment analysis was performed by human reviewers, ChatGPT, and Claude AI. High-engagement videos were defined as those within the upper quartile for likes (≥30,400) or shares (≥1,050).
    RESULTS: Most videos were created by patients (89%) and focused on postoperative experiences (75%). Videos demonstrated substantial engagement, with median values of 3,860.5 likes, 131 comments, and 194.5 shares. Overall informational quality was poor, with median GQS and mDISCERN scores of 1-2 across raters. Highly liked and highly shared videos were significantly shorter in duration (18 vs. 34 seconds; p < 0.05) and generated markedly greater engagement across all metrics. Popular videos received significantly lower quality scores from human reviewers. Sentiment analysis revealed predominantly positive responses (49-58%). Claude showed greater concordance with human consensus than ChatGPT (κ = 0.749 vs. 0.649).
    CONCLUSION: Breast reduction content on TikTok is largely patient-driven and characterized by high engagement but low informational quality. Shorter videos achieved greater dissemination despite being associated with less reliable health information.
    Keywords:  breast surgery; communication; mammaplasty; plastic and reconstructive surgery; social media
    DOI:  https://doi.org/10.7759/cureus.117785
  19. J Cancer Surviv. 2026 Nov 11.
       PURPOSE: Tamoxifen, the first targeted anticancer agent for breast cancer, reduces recurrence and mortality. However, nonadherence remains a significant barrier. We investigated the quality, readability, distractibility, and tone of online information regarding tamoxifen.
    METHODS: We identified the top four search terms from 2020 to 2025 for tamoxifen using Google Trends. These terms were queried by two independent researchers, and re-queried 2 weeks later to maximize cross-sectional accuracy. Final websites were categorized by authorship. Quality was assessed using the DISCERN instrument and JAMA benchmark criteria. Readability was scored using the Flesch Reading Ease, Flesch-Kincaid Grade Level, Coleman-Liau Index, and SMOG index formula. Distractibility was measured using a validated tool created by Truumees et al. Tone was measured with IBM Watson Natural Language Understanding.
    RESULTS: Fifty-five unique websites were included. Journal repositories scored significantly higher in quality compared to social media websites (p < 0.05), which had the lowest quality of information. The average text complexity ranged at or above the twelfth grade reading level. Higher quality positively correlated with higher reading level (p < 0.05). There was no difference in the distractibility index or tones across website categories.
    CONCLUSIONS: Online information regarding tamoxifen is overall negative, varies in quality, and may be difficult to understand. Higher quality websites are most difficult for patients to understand, which is detrimental to informed decision-making. Implications for Cancer Survivors Cancer Survivors increasingly access online information. This study provides insight into the websites encountered when searching for tamoxifen online. Clinicians can utilize this information to guide conversations.
    IMPLICATIONS FOR CANCER SURVIVORS: Cancer Survivors increasingly access online information. This study provides insight into the websites encountered when searching for tamoxifen online. Clinicians can utilize this information to guide conversations.
    Keywords:  Online information; Patient education; Quality; Readability; Tamoxifen; Tone
    DOI:  https://doi.org/10.1007/s11764-026-02155-6
  20. Brain Spine. 2026 ;6 107718
       Introduction: Patients increasingly seek information about complex neurosurgical conditions online. The quality of such resources has not been systematically studied for vascular neurosurgery and skull base surgery.
    Research question: How do German- and English-language online resources for eight vascular and skull base neurosurgical conditions compare in reliability and transparency, and is search rank associated with quality?
    Material and methods: A simulated Google search was performed for nine search terms covering eight conditions in German and English. The first 20 results per search were rated by two independent raters using the modified DISCERN Plus score (16-80) and JAMA benchmarks (0-4). Non-parametric tests, effect sizes and intraclass correlation coefficients were calculated.
    Results: In total, 355 websites were included (177 English, 178 German). Mean DISCERN score was 41.1 ± 9.7 and mean JAMA score 2.61 ± 0.94 (fair quality). English-language sites scored higher than German sites on both instruments, with small effect sizes (r = 0.12-0.13). Host type was strongly associated with JAMA scores (ε2 = 0.25); educational and governmental sites scored highest. Among English-language results, top-ranked websites were of higher quality (rho = -0.315; positions 1-5: median DISCERN 49.8 vs. 40.0-42.0 thereafter), whereas rank was unrelated to quality in German-language results. Interrater agreement was good to excellent (ICC 0.79-0.82).
    Discussion and conclusion: Online resources for these conditions are generally of fair quality. Top-ranked English-language results tended to be of higher quality, whereas search rank was not informative for German-language results. Guiding patients, particularly German-speaking patients, towards reliable resources remains essential.
    Keywords:  DISCERN; JAMA benchmarks; Online health information; Patient education; Skull base surgery; Vascular neurosurgery
    DOI:  https://doi.org/10.1016/j.bas.2026.107718
  21. Patient Educ Couns. 2026 Nov 06. pii: S0738-3991(26)00914-6. [Epub ahead of print]154 110381
       OBJECTIVES: This study evaluated the content and readability of online health information about selective dorsal rhizotomy (SDR) for children with cerebral palsy (CP) from internationally ranked pediatric hospital websites, and assessed alignment with shared decision-making principles.
    METHODS: We analyzed English-language (or translated) webpages from Newsweek's top pediatric hospital websites (2023-2025). Content was assessed using a predefined framework, readability via the Simple Measure of Gobbledygook, and decision aid qualifying criteria using the International Patient Decision Aid Standards.
    RESULTS: Of 284 screened webpages, 67 met inclusion criteria. Most were from United States (63%) and in English (93%). Indications identified as being appropriate for SDR were more likely to be described (86%) compared to indications not suitable (16%). Benefits were reported more often than harms (85% vs 22%). Most webpages reported procedural details (76%), with 43% reporting general rehabilitation information. Fewer webpages included patient experience testimonies (16%), references (15%) and financial considerations (9%). Mean readability score was grade 13.9 compared to the recommended grade 6-8 level, and over half (54%) had more than 30% complex language, in contrast to health literacy guidelines. Decision aid qualifying criteria were rarely met, with only 58% including a description of the health condition.
    CONCLUSION: Freely accessible online information about SDR is often imbalanced, and written above recommended readability levels, potentially creating barriers to understanding and limiting its capacity to support informed decision-making.
    PRACTICE IMPLICATIONS: Accessible and easy-to-read information may help families engage in discussions with healthcare professionals and participate more meaningfully in decisions about SDR. Healthcare organisations may improve the balance, readability, and decision-support quality of digital SDR resources. Presenting clear information on benefits and harms, suitability and non-suitability criteria, rehabilitation demands, costs, and lived experiences, and embedding shared decision-making principles through evidence-based, co-designed resources may enhance family understanding and better support clinician-family discussions about SDR.
    Keywords:  Cerebral palsy; Decision-making; Online information resources; Patient education; Pediatrics; Selective dorsal rhizotomy
    DOI:  https://doi.org/10.1016/j.pec.2026.110381
  22. NPJ Digit Med. 2026 Jul 23. pii: 887. [Epub ahead of print]9(1):
      YouTube is a primary health information source, yet its engagement-driven model may incentivize "clickbait," potentially compromising quality. We analyzed clickbait prevalence in 3958 Korean medical YouTube videos across five health topics and its relationship with engagement, quality, and source. Videos were assessed for clickbait severity (Clickbait Severity Score; CSS), engagement, and medical information quality using a composite score (Medical Quality Score; MQS) derived from Journal of the American Medical Association Benchmark (credibility/accountability), DISCERN (reliability), and Global Quality Scale (overall quality) criteria. Independent YouTubers produced the lowest quality content (MQS mean = -0.3) with the most clickbait (CSS mean = 1.4), yet garnered the highest viewership. Clickbait severity was positively correlated with log-transformed daily views (r = .231, p < .001) and negatively correlated with medical quality (r = -0.371, p < 0.001). Multiple regression analysis revealed that specific clickbait tactics like exaggeration (β = 0.161, p < 0.001) and teasing (β = 0.068, p < 0.001) were positive predictors of daily views, whereas higher information quality (GQS) was a negative predictor (β = -0.185, p < 0.001). These findings demonstrate a "credibility paradox" where algorithms reward clickbait with higher engagement despite lower content quality. This misalignment with public health principles highlights an urgent need for platform governance reform to better align algorithmic incentives with information quality.
    DOI:  https://doi.org/10.1038/s41746-026-03046-x
  23. J Paediatr Child Health. 2026 Nov 10.
       AIM: YouTube is frequently used by parents as a primary source of information for paediatric emergencies, including infant choking. This study aimed to evaluate the comprehensiveness, reliability, and quality of YouTube videos regarding choking management in infants (0-12 months) and to assess whether video popularity is associated with content quality.
    METHODS: Of the first 200 videos retrieved by a YouTube search on January 20, 2026, 40 met predefined eligibility criteria and were analysed using the modified DISCERN (mDISCERN), the Global Quality Scale (GQS), and a guideline-derived Infant Choking Management Checklist (ICMC; 0-20). Kruskal-Wallis tests and Spearman correlations with false discovery rate adjustment were used.
    RESULTS: Educational health channels (40.0%) and official health institutions (22.5%) were the primary uploaders. The median ICMC score was 14.0 (IQR 12.0-15.5), and median GQS and mDISCERN scores were both 3.0 (moderate quality and reliability). Intervention technique steps were almost always demonstrated, whereas recognition and safety steps were frequently omitted; only 18.8% of videos warned against blind finger sweeps. Scores did not differ significantly by video source or practitioner background (p > 0.05). ICMC scores correlated strongly with GQS scores (ρ = 0.834, p < 0.001), whereas engagement metrics were not associated with any content score.
    CONCLUSIONS: In this sample, YouTube videos on infant choking management were of moderate quality and reliability and often omitted the recognition and safety steps of the guideline algorithm. No association between popularity and content quality was identified. Healthcare professionals should guide parents towards vetted, evidence-based resources covering the complete algorithm.
    Keywords:  YouTube; airway obstruction; choking; foreign body; infant
    DOI:  https://doi.org/10.1111/jpc.70652
  24. J Hand Ther. 2026 Nov 11. pii: S0894-1130(26)00113-4. [Epub ahead of print]
       BACKGROUND: YouTube is increasingly used by patients seeking information about De Quervain's tenosynovitis; however, the quality, reliability, and educational completeness of the available videos remain uncertain.
    AIM: The aim of this study is to evaluate the quality, reliability, and educational content of YouTube videos related to De Quervain's tenosynovitis.
    STUDY DESIGN: A cross-sectional study.
    METHODS: A YouTube search was conducted using the terms "De Quervain's Tenosynovitis," "De Quervain's Tenosynovitis Exercise," and "Mommy's Thumb". The keywords were selected by reviewing previous studies and the top 50 search results for each keyword were evaluated because this method was frequently encountered in other studies. The Journal of the American Medical Association Benchmark Criteria (JAMA-BC) was used to assess each video's reliability, and the Global Quality Score (GQS), the Modified DISCERN (mDISCERN) instrument, and the De Quervain's tenosynovitis-specific score (DQT-SS) were used to assess its quality.
    RESULTS: When the sources of the videos were evaluated, 33.3% were uploaded by physicians, 47.1% by nonphysician healthcare professionals, and 19.6% by official institutional channels. According to the JAMA-BC, 9.8% of the videos were classified as low reliability, 72.5% as medium reliability, and 17.6% as high reliability. The GQS showed that 51.0% of the videos were low quality, 37.3% medium, and 11.8% high. Similarly, according to the mDISCERN, 52.0% reported low confidence, 38.0% reported medium confidence, and 10.0% reported high confidence. According to the DQT-SS, 25.5% of the videos had low, 60.8% had medium, and 13.7% had high educational quality.
    CONCLUSION: Healthcare professionals should consider using standardized assessment tools such as the JAMA-BC and the DQT-SS when recommending YouTube videos to patients. Patients should preferentially use videos produced by identifiable healthcare professionals that provide transparent and evidence-based information.
    Keywords:  De Quervain’s tenosynovitis; Mommy’s thumb; Video; YouTube
    DOI:  https://doi.org/10.1016/j.jht.2026.09.012
  25. JMIR Dermatol. 2026 Nov 09. 9 e88827
       Background: The era of social media platforms has transformed the dissemination of health information, raising concerns about the quality and accuracy of content. Urticaria, a common skin condition, necessitates reliable information for effective management. Unreliable information on the internet can lead to misconceptions about urticaria, which could contribute to overemphasis on triggers or underrepresentation of guideline-based interventions, delaying effective management approaches.
    Objective: This study aimed to systematically evaluate urticaria-related content on 2 prominent social media platforms, YouTube and TikTok, using validated tools to assess information reliability and explore the characteristics of videos and their uploaders.
    Methods: On February 16, 2023, a comprehensive search on YouTube and TikTok was conducted for videos related to urticaria. Two independent dermatologists rated 92 eligible videos using DISCERN, the Global Quality Score (GQS), and the Accuracy in Digital-Health Instrument (ANDI). Statistical analyses, including mean values, SDs, percentages, and chi-square tests, were used to interpret the data.
    Results: TikTok videos were predominantly uploaded by patients (29/46, 63%), whereas YouTube exhibited a higher engagement of health care professionals (31/46, 67.4%). YouTube videos exhibited higher mean scores for the GQS (2.9, SD 0.7), ANDI (2.3, SD 0.8), and DISCERN (40.2, SD 7.6) compared to TikTok (GQS: mean 2, SD 0.8; ANDI: mean 1.4, SD 1; DISCERN: mean 32.9, SD 7.9). Videos uploaded by professionals on TikTok had significantly higher mean GQS (2.8, SD 0.7), ANDI (2.1, SD 0.8), and DISCERN (40, SD 6.3) scores compared to those uploaded by nonprofessionals (P=.001). Similarly, videos from professional uploaders on YouTube exhibited significantly higher mean GQS (3, SD 0.6), ANDI (2.6, SD 0.6), and DISCERN (42.2, SD 7.1) scores compared to nonprofessionals (P=.01 for GQS, P=.001 for ANDI, and P=.003 for DISCERN).
    Conclusions: Urticaria videos on TikTok showed a significantly higher engagement ratio and less accuracy and quality of health-related information than YouTube videos. These insights can guide content creators, platform administrators, and health care professionals in collaboratively fostering a trustworthy and accurate health information environment on social media.
    Keywords:  ANDI; Accuracy in Digital-Health Instrument; DISCERN; GQS; Global Quality Score; TikTok; YouTube; content quality; health care professionals; patient uploaders; social media; urticaria
    DOI:  https://doi.org/10.2196/88827
  26. J Robot Surg. 2026 Nov 10. pii: 1195. [Epub ahead of print]20(1):
      Surgeons in training increasingly use YouTube to prepare for partial nephrectomy (PN), yet how completely these videos report procedure-specific information, and whether this differs by surgical platform, is unknown. On 6 August 2026, the first 250 results of two YouTube searches ("robotic partial nephrectomy" and "laparoscopic partial nephrectomy") were retrieved. Two raters independently scored 248 eligible videos from 130 channels with an adapted 16-item LAP-VEGaS checklist using PN-specific item definitions, together with the JAMA benchmark criteria, modified DISCERN and the Global Quality Score. Regression models accounted for clustering of videos within channels. Of 248 videos, 139 were robotic, 109 laparoscopic and 67 (27.0%) institutional. Institutional origin was associated with higher scores on all four instruments in adjusted models. Technical items were almost always met (stepwise demonstration 99.6%), whereas pathology and margin status (25.4%) and ischaemia management (42.7%) were seldom reported. Robotic videos reported ischaemia management more often overall (52.5% vs. 30.3%), but not within individual channels (30.4% vs. 27.5%; p = 0.60), because 60 of the 67 institutional videos were robotic. Margin reporting was the only item consistently associated with the robotic approach (adjusted odds ratio 3.46, 95% CI 1.16-10.38), although the interval included 1 under the strictest rating definition. Popularity was a poor guide to quality. Online PN videos reliably demonstrate technique but seldom report the information needed to appraise its outcome, and much of the apparent superiority of robotic content reflects its institutional source. A PN-specific minimum reporting set, to be developed by expert consensus, is proposed.
    Keywords:  LAP-VEGaS; Partial nephrectomy; Reporting quality; Robotic surgery; Surgical education; YouTube
    DOI:  https://doi.org/10.1007/s11701-026-04185-9
  27. BMC Oral Health. 2026 Aug 12. pii: 2208. [Epub ahead of print]26(1):
       BACKGROUND: YouTube is increasingly used as an accessible source of educational information in dentistry; however, the quality, reliability, and instructional value of videos on dental photography training remain unclear. This study aimed to evaluate the educational quality, reliability, and engagement characteristics of YouTube videos related to dental photography.
    METHODS: A cross-sectional analysis was conducted using the search terms "Dental Photography" and "Mobile Dental Photography" on YouTube videos. The top 100 videos for each keyword were screened, and 113 eligible videos were included. Video characteristics such as duration, view count, likes, dislikes, view rate, and engagement index were recorded. Video quality and reliability were assessed using the Modified Global Quality Scale (mGQS), the DISCERN tool, and the Journal of the American Medical Association (JAMA) benchmark criteria. Videos were categorised by upload source as dentist/clinician, academic institution, or health website/organisation. Statistical analyses were performed using the Kruskal-Wallis and Spearman correlation tests (p < 0.05).
    RESULTS: The mean mGQS, DISCERN, and JAMA scores were 2.18 ± 0.91, 30.27 ± 8.26, and 1.34 ± 0.82, respectively, indicating generally low quality and reliability. Videos uploaded by dentists/clinicians had significantly higher mGQS, DISCERN, and JAMA scores compared to those uploaded by health websites (p < 0.05). Academic videos also showed relatively higher quality scores, despite their limited number. Significant positive correlations were found among the quality scores.
    CONCLUSIONS: Videos on dental photography posted on YouTube have generally been of low quality and reliability. Videos uploaded by dentists/clinicians and academic sources tended to provide higher-quality information compared to those uploaded by health websites. With greater involvement from dental professionals and academic institutions, this topic will become even more significant.
    Keywords:  Dental photography; Digital dentistry; Internet; Mobile dental photography; Social media; Youtube
    DOI:  https://doi.org/10.1186/s12903-026-09591-x
  28. Obes Surg. 2026 Nov 10.
       BACKGROUND: Social media is an important source of health information for patients considering metabolic and bariatric surgery (MBS), but the quality of this information remains unclear. We assessed the quality of MBS-related videos on TikTok, Instagram, and Facebook.
    METHODS: This cross-sectional study evaluated English-language videos identified between October and December 2025 using four MBS-related hashtags. Two MBS-experienced reviewers independently assessed video quality using the validated DISCERN and Principles for Health-related Information on Social Media (PRHISM) instruments. Video characteristics, creator type, and engagement metrics were recorded and compared across platforms and creator groups.
    RESULTS: Of 116 included videos, 55 were from TikTok, 28 from Instagram and 33 from Facebook. Overall information quality was low to moderate. Instagram had the highest DISCERN and PRHISM scores, followed by Facebook, while TikTok had the lowest. In unadjusted analyses, DISCERN scores were significantly lower on TikTok than Instagram, while PRHISM scores were significantly lower on TikTok than both Instagram and Facebook. Healthcare professionals produced videos with significantly higher DISCERN scores than non-medical creators (34.05 ± 6.63 vs 25.66 ± 5.77; p < 0.001) and higher PRHISM scores (49.14 ± 9.16 vs 31.53 ± 7.83; p < 0.001). However, non-medical creators received significantly more likes and shares. Greater engagement was associated with lower information quality in unadjusted analyses, but these associations were attenuated after adjustment for platform and creator type.
    CONCLUSION: The quality of MBS-related information in the sampled social-media videos was low to moderate. Although TikTok showed lower information-quality scores than Instagram and Facebook in unadjusted analyses, the platform difference in DISCERN scores was attenuated after adjustment for creator type. Healthcare professionals provided higher-quality content but received less engagement, highlighting the need for accurate, accessible, and engaging evidence-based MBS information on social media.
    Keywords:  DISCERN · TikTok; Health information quality; Metabolic and bariatric surgery; Patient education; Social media
    DOI:  https://doi.org/10.1007/s11695-026-09035-z
  29. JMIR Infodemiology. 2026 Nov 12. 6 e91852
       BACKGROUND: Breast cancer screening plays a critical role in early detection and mortality reduction. Short video platforms have become widely used channels for patient education, but the quality and reliability of breast cancer screening information on short video platforms remain unclear.
    OBJECTIVE: This study aimed to explore and compare the source, content characteristics, and quality of short videos on breast cancer screening across 2 major Chinese-language platforms (Douyin and Kuaishou) and the English-language platform TikTok.
    METHODS: Using the keyword "breast cancer screening," this study retrieved and screened the top 100 short videos from each of the 3 platforms-Douyin, Kuaishou, and TikTok-resulting in a final sample of 258 eligible videos. Basic video information, user engagement metrics, and uploader characteristics were collected. A total of 3 breast disease specialists independently evaluated the content using 4 tools: the Patient Education Materials Assessment Tool for Audiovisual Materials (PEMAT-A/V), modified DISCERN (mDISCERN), Global Quality Score (GQS), and Journal of the American Medical Association benchmark criteria (JAMA-BC), which assess understandability and actionability, overall quality, educational value, and reliability, respectively. Statistical analyses included Kruskal-Wallis tests, Dunn-Bonferroni tests, chi-square tests, and Spearman correlation analysis.
    RESULTS: Videos on TikTok had significantly lower scores on the mDISCERN, GQS, and JAMA-BC than those on Douyin and Kuaishou, indicating poorer overall quality and reliability. The median mDISCERN scores were 1.50 (IQR 1.00-2.00) for TikTok compared to 3.00 (IQR 2.00-3.00) for both Douyin and Kuaishou (P<.001), the median GQS scores were 2.00 (IQR 2.00-3.00) for TikTok compared to 3.00 (IQR 2.50-4.00) (P<.001) for Douyin and 3.00 (IQR 3.00-3.50) (P<.001) for Kuaishou, and the median JAMA-BC scores were 10.00 (IQR 8.25-12.00) for TikTok compared to 12.00 (IQR 10.00-12.00) (P<.001) for Douyin and 12.00 (IQR 10.00-14.00) (P<.001) for Kuaishou. The PEMAT-A scores of Douyin and Kuaishou were also significantly lower on TikTok (Douyin vs TikTok: P=.002; Kuaishou vs TikTok: P=.02). Health care professionals were the main content creators on Douyin (49/66, 74.2%) and Kuaishou (56/69, 81.1%), while individual content creators dominated TikTok (33/72, 45.8%). Authentication rates were significantly lower on TikTok (9/72, 12.5%) than on Kuaishou (35/69, 50.7%) and Douyin (58/66, 87.9%). Viewer engagement metrics, including likes, comments, and collections, were nonsignificant or negatively correlated with video quality. For example, mDISCERN scores were negatively correlated with the number of likes per 30 days (P<.001).
    CONCLUSIONS: This study reveals that high-quality videos on breast cancer screening tend to attract lower viewer engagement, highlighting platform-specific disparities and challenges in dissemination. It is recommended that short video platforms implement stricter review processes for health-related videos and uploaders. Additionally, governmental or related agencies should provide media training for health care professionals in public science communication to enhance the impact and effectiveness of health education via short videos.
    Keywords:  breast cancer screening; health communication; health education; short video; video quality
    DOI:  https://doi.org/10.2196/91852
  30. Therapie. 2026 Oct 13. pii: S0040-5957(26)00141-1. [Epub ahead of print]
       AIMS: To assess the quality and reliability of information about antidepressant-related content on the social media platforms Instagram and TikTok, two platforms widely used by adolescents and young adults.
    METHODS: A cross-sectional analysis of 200 French-language posts (100 from each platform) was conducted using the DISCERN scale, along with additional measures for accuracy, promotion of adherence and audience. The evaluations were performed by three independent evaluators. Descriptive and univariate analyses were carried out to compare the characteristics of information between the two media platforms according to author qualification.
    RESULTS: The overall quality was rated as "fair" (DISCERN score=32.0±10.5) with health professionals providing significantly higher quality, more reliable, more accurate, and more likely to promote adherence than non-health professionals. Overall, 23.5% of posts contained misinformation and 33.5% discouraged treatment adherence. The main topics addressed were the effectiveness of antidepressants (27.3%) and their adverse effects (23.9%), with the most common being weight gain, withdrawal syndrome, and sexual dysfunction.
    CONCLUSION: This exploratory study suggests an unsatisfactory quality of information about antidepressants available on Instagram and TikTok, with a worrying proportion of misinformation. These findings highlight the importance of developing initiatives to improve the visibility and dissemination of reliable, evidence-based health information on social media.
    Keywords:  Adolescent; Antidepressant drug; Mental health; Social media; Young adult
    DOI:  https://doi.org/10.1016/j.therap.2026.09.048
  31. AIDS Behav. 2026 Nov 09.
      Online health information seeking can provide people living with HIV/AIDS (PLWHA) with discreet access to illness-related information, but its use is shaped by stigma, privacy concerns, trust, and practical constraints. This descriptive phenomenological study explored the lived experiences and behavioral logic of online health information seeking among 20 PLWHA recruited from three designated HIV/AIDS care institutions in southwestern China. Semi-structured interviews were analyzed using Colaizzi's seven-step method. Four themes were identified: seeking information to manage treatment, interpret symptoms, understand the illness, and sustain hope; seeking online health information while protecting privacy through socially enabled but often concealed searching; judging sources according to accessibility, authority, comprehensibility, and timely interaction; and encountering mismatch, information overload, negative content, and situational barriers that reduced usefulness. Participants moved between active searching, cautious searching, limiting searches, and selectively avoiding further engagement according to privacy risk, anticipated judgment, emotional burden, perceived relevance, and digital capacity. Online information was valued for convenience and lower visibility, whereas healthcare professionals remained important for individualized interpretation and clinical decisions. Online health information seeking among PLWHA is therefore better understood as a privacy-sensitive process of digital health navigation rather than simple information acquisition. Digital health support should provide discreet access, trustworthy and comprehensible content, responsive clarification, and pathways suited to different information needs and levels of digital capacity.
    Keywords:  Digital health navigation; Online health information seeking; PLWHA; Privacy protection; Qualitative research; Stigma
    DOI:  https://doi.org/10.1007/s10461-026-05331-y
  32. Arthrosc Sports Med Rehabil. 2026 Nov 09. e70084
       Purpose: To identify the common patient questions and to evaluate the quality of online resources regarding peptide therapy, namely, body protection compound-157, in orthopaedic surgery.
    Methods: Three orthopaedic peptide searches were entered into a clean Google Web Search on June 30, 2025. Questions under the "People also ask" tab were expanded sequentially. The first 100 consecutive results meeting inclusion criteria for each query were included (300 total). Questions or websites unrelated to orthopaedic peptides and repeat verbatim questions within each query were excluded. Questions were categorized into 1 of 3 Rothwell Classifications and subcategorized by topic. Websites were categorized based on source. The Journal of the American Medical Association Benchmark Criteria was used to evaluate the quality of each website.
    Results: Most questions fell into the Rothwell Fact category (61.0%). Most common question topics were Structure/Function (31.7%), Risks/Complications (20.7%), and Treatment Comparisons (9.7%). The least common question topics were Specific Activities (0%), Pain (0.3%), Longevity of Treatment (0.7%), and Management (0.7%). The most common websites were Commercial (28.0%), Academic (18.0%), and Government (18.0%). PubMed articles consisted of 98.1% of Government websites. The overall average Journal of the American Medical Association score was 2.0 ± 1.5. Government websites had the highest score (3.6 ± 0.7) and constituted 48.6% of websites with a score of 4/4. Medical Practice and Single Surgeon websites had the lowest Journal of the American Medical Association score of 0.7 ± 1.0 and 0.7 ± 0.9, respectively.
    Conclusions: The most common online patient questions pertain to the structure, function, and risks associated with peptide therapy. The quality of the information is heterogeneous.
    Clinical Relevance: Because of the increasing popularity and accessibility, improved technology, and limited research, physicians should be aware of the most common online patient questions pertaining to peptide therapy in orthopaedics. Care should be taken when presenting the impact of peptides without discussing the quality to avoid overvaluing its impact.
    DOI:  https://doi.org/10.1002/ars2.70084
  33. Healthcare (Basel). 2026 Oct 22. pii: 3558. [Epub ahead of print]14(21):
       BACKGROUND/OBJECTIVES: Saudi Health consumers increasingly rely on generative Artificial Intelligence (AI) to access health information from a growing number of health-related resources. Little is known about the frequency and patterns of generative AI tool use for seeking health information among consumers, its reported impact, and health consumers' perceptions of these tools. This study aims to investigate the use of and perspectives on generative AI tools among health consumers in Saudi Arabia (SA). It also aims to assess the impact of their use on health-seeking behaviours.
    METHODS: An exploratory, cross-sectional, self-administered online survey was distributed to adults (≥18 years) residing in SA during two data collection rounds in 2026 (March-April and July-August). All adults who used generative AI tools to seek online health information in the past year were included. Bivariate and multivariate analyses were conducted to examine relationships among variables.
    RESULTS: A total of 403 eligible health consumers (58% male) participated. Among respondents who had used generative AI for health purposes, approximately 49% reported using it more than five times per month. Younger age, higher income, and more frequent pre-use were associated with higher odds of frequent seeking health information behaviour. Headache, abdominal pain, and musculoskeletal pain were the most common claims. ChatGPT was the main platform (84%) used for seeking information. Age was significantly negatively associated with frequency, whereas income and pre-visit use were positively associated. The primary reasons for seeking health information via generative AI were quick information retrieval (71%), time saving (34%), and curiosity (30.5%).
    CONCLUSIONS: The high frequency of use suggests that generative AI can serve as a source of health information and may lead users to withhold health information from healthcare providers. It may influence subsequent health-seeking behaviours, such as requesting urgent care or postponing physician visits. These self-reported findings do not establish diagnostic accuracy, clinical effectiveness, or patient safety, and the clinical risk and appropriateness of self-management actions were not assessed. Many respondents reported concerns about inaccurate information, missed serious illness, and increased health anxiety.
    Keywords:  Saudi Arabia; artificial intelligence; consumer health informatics; health consumer; online health information
    DOI:  https://doi.org/10.3390/healthcare14213558
  34. BMC Public Health. 2026 Aug 12. pii: 3260. [Epub ahead of print]26(1):
       BACKGROUND: Adults living with chronic disease increasingly depend on online sources for day-to-day self-management, yet digital environments expose them to health misinformation of variable credibility. Prior research has largely examined information seeking, verification, or misinformation exposure in isolation, leaving the behavioural mechanisms that connect them poorly understood.
    OBJECTIVE: Drawing on dual-process accounts of information processing, this study examined how online health information seeking (HIS), verification behaviour (VER), and misinformation-related perceptions (MIS) interrelate among adults with chronic disease and tested whether verification mediates the association between information engagement and misinformation awareness.
    METHODS: A cross-sectional online survey was completed by 740 adults with self-reported chronic disease, recruited through messaging platforms and patient support organisations in Saudi Arabia. Validated five-point Likert-scale items measured HIS, VER, and MIS. Data were analysed using descriptive statistics, visual analytics, and structural equation modelling (SEM) in R (lavaan); construct reliability and convergent validity were assessed via composite reliability and average variance extracted. Descriptive analyses were based on all 740 respondents (item-level n = 724-740, as most items were optional), whereas the structural model used the 680 cases with complete data on all modelled items.
    RESULTS: Participants reported frequent online health information seeking and widespread use of verification strategies, alongside high awareness of misleading social media content. On re-estimation using the final verified dataset, global model fit was below conventional thresholds (CFI = 0.874; TLI = 0.844; RMSEA = 0.153; SRMR = 0.090), so the structural estimates are interpreted descriptively and with caution. HIS was strongly and positively associated with VER (β = 0.82, P < .001). VER was positively associated with MIS (β = 0.73, P < .001), a pattern that may reflect heightened evaluative awareness of misleading content rather than passive susceptibility, although this interpretation remains tentative given the cross-sectional design and the high correlations among constructs. The direct HIS→MIS path was non-significant (β = 0.09, P = .21), indicating that information engagement related to misinformation perceptions primarily through verification.
    CONCLUSIONS: Findings are consistent with a behaviourally integrated, mediated model in which online information seeking is associated with greater verification, and verification, as a form of systematic evaluative processing, is associated with greater awareness of misinformation rather than vulnerability to it; because the data are cross-sectional and the constructs overlap empirically, these pathways describe associations rather than demonstrated causal effects. By modelling these constructs simultaneously, the study moves beyond descriptive accounts and positions chronic disease patients as active appraisers of digital health information. Interventions that reinforce verification skills and embed credibility cues within commonly used platforms may strengthen safe digital health engagement.
    Keywords:  Chronic disease self-management; Digital health literacy; Health information verification; Health misinformation; Online health information seeking; Patient information behaviour; Saudi Arabia; Structural equation modelling
    DOI:  https://doi.org/10.1186/s12889-026-28721-2