bims-arines Biomed News
on AI in evidence synthesis
Issue of 2026–07–05
twenty papers selected by
Farhad Shokraneh, Systematic Review Consultants LTD



  1. medRxiv. 2026 Jun 24. pii: 2026.06.13.26355589. [Epub ahead of print]
       Background: Systematic reviews of observational studies are central to causal inference in chronic disease epidemiology but are increasingly limited by the scale of the literature and heterogeneity in confounder control. There is a need for transparent, open methods that reduce screening burden and make reported exposures, outcomes, and covariates comparable across studies.
    Objective: To develop and evaluate modular LLM-based pipelines, LitScreen and VarEx, that automate study screening and variable extraction for observational systematic reviews across multiple use cases, including hypertension as a primary exposure with Alzheimer's disease and related dementias (ADRD) as outcomes, and posttraumatic stress disorder (PTSD) as the exposure with self-harm, self-injury, and suicidality as outcomes.
    Methods and Materials: We built an end-to-end workflow in which reproducible MEDLINE via Ovid queries yield RIS corpora that are processed by LitScreen, a three-phase screening pipeline combining abstract-level evidence extraction, criterion-wise inclusion adjudication with high-recall gates, and full-text retrieval-augmented verification. Screened-in articles enter VarEx, a retrieval-augmented extraction pipeline that identifies role-specific passages and performs evidence-grounded extraction and semantic classification of exposures, outcomes, and covariates into predefined categories aligned with Metaconfoundr. Performance was evaluated on six labeled SYNERGY datasets and expert-annotated hypertension-to-ADRD and education-to-dementia corpora using precision, recall, F 1 , a strict score requiring correct variable identity and category, and time-per-reference estimates.
    Results: In the primary hypertension-to-ADRD reference set, VarEx achieved covariate-level precision of 0.80, recall of 0.79, and F 1 of 0.76, with classification accuracy of 0.97 and similar performance for education-to-dementia and SYNERGY validation datasets. LitScreen preserved high recall while excluding most ineligible records and reduced total screening and extraction time by roughly 80-90 percent relative to manual review baselines by routing only uncertain or borderline citations to full-text verification.
    Conclusion: A retrieval-augmented LLM framework can automate major components of screening and variable extraction for observational systematic reviews, generating reusable structured covariate inventories that integrate with causal confounder assessment tools and substantially improve the efficiency and reproducibility of evidence synthesis, while remaining an assistant to, rather than a replacement for, human reviewers.
    DOI:  https://doi.org/10.64898/2026.06.13.26355589
  2. Stud Health Technol Inform. 2026 Jun 29. 338 130-134
      Language bias arises in systematic reviews when non-English studies are excluded owing to resource constraints. Large language models (LLMs) can mitigate this problem through multilingual processing. To assess whether direct multilingual LLM processing reduces language-based disparities in systematic review screening performance compared to translation-mediated approaches. Six state-of-the-art LLMs were evaluated under three conditions: (1) an English benchmark dataset (n = 2,911), (2) direct screening of non-English abstracts (n = 483), and (3) screening of machine-translated non-English abstracts. Performance was measured using sensitivity, specificity, F1 score, balanced accuracy, and workload reduction. All models achieved high sensitivity on English data (≥0.938). Translation-mediated screening substantially reduced sensitivity in some models (range: 0.47-0.54), whereas direct multilingual processing maintained high sensitivity (range: 0.71-1.00). Considerable differences were observed among models. Direct multilingual LLM screening may reduce language-related sensitivity disparities; however, the effects on downstream meta-analytic bias require further investigation.
    Keywords:  Large language models; language bias; screening automation; systematic review
    DOI:  https://doi.org/10.3233/SHTI260813
  3. medRxiv. 2026 Jun 26. pii: 2026.06.16.26355773. [Epub ahead of print]
       Background: Large language models (LLMs) offer promise for systematic review data extraction, but performance in complex multidisciplinary domains and utility for clinical statement generation remain insufficiently described.
    Objectives: To evaluate Google NotebookLM for AI-assisted data extraction and RAND/UCLA consensus statement generation in a systematic review of IBD, obesity, and cardiometabolic comorbidities.
    Methods: Studies were organized into domain-specific notebooks; structured prompts generated standardized evidence tables. Two independent reviewers validated outputs against full-text articles using a four-category error classification. Cell-level accuracy and critical accuracy (cells free of major factual errors) were the primary metrics; workflow time was compared against a published conventional extraction benchmark. Concordance between AI-generated and expert-finalized statements was assessed.
    Results: Across 57 articles, 1,710 data cells were extracted; 151 (8.83%) were flagged, yielding 91.17% cell-level accuracy. Major factual errors occurred in only 4 cells (0.23%), for a critical accuracy of 99.77%. Most errors were minor omissions (59.6%) or incomplete extractions (30.5%); domain error rates ranged from 7.08% to 11.33%. The pipeline required 17.7 versus a projected 165.1 person-hours (89.3% reduction). PICO-structured prompting generated 70 candidate statements; 58 of 112 finalized panel statements (51.8%) were AI-derived, and 85.7% were retained in the finalized set.
    Conclusion: Google NotebookLM demonstrates feasibility as a primary extraction and synthesis tool in a multidisciplinary systematic review, with extractive incompleteness as the principal limitation and substantial time savings over conventional approaches. Its novel application to RAND/UCLA consensus statement generation extends AI-assisted evidence synthesis to clinical consensus generation workflow.
    DOI:  https://doi.org/10.64898/2026.06.16.26355773
  4. Int J Med Inform. 2026 Jun 26. pii: S1386-5056(26)00301-1. [Epub ahead of print]219 106561
       BACKGROUND: Evaluating adherence to PRISMA 2020 guideline remains a burden in the peer review process. However, there is a lack of shareable benchmarks for evaluating large language model (LLM) performance in this task.
    METHODS: We constructed a copyright-aware benchmark of 108 Creative Commons-licensed systematic reviews. We first conducted parameter optimization using five SRs from the Suda dataset, then compared five checklist input formats (Markdown, JSON, XML, plain text, and manuscript-only control) using ten development-phase LLMs on ten further SRs from the Suda dataset, and finally validated the locked Markdown pipeline using nineteen LLMs on ten SRs from the Tsuge dataset as additional frontier models became available during the study period.
    RESULTS: Supplying structured PRISMA 2020 checklists yielded 78.7-79.7% accuracy versus 45.2% for manuscript-only input, with paired aggregate analyses showing that structured formats outperformed manuscript-only input while structured formats did not differ significantly from one another. In the validation sample, accuracy ranged from 68.5% to 86.0% with distinct sensitivity-specificity trade-offs. Using Qwen3-Max on the full dataset (n = 120), we achieved 95.1% sensitivity and 49.3% specificity.
    DISCUSSION: Structured checklist provision substantially improves LLM-based PRISMA assessment. However, given the observed proportion of false positives, human expert verification remains essential before editorial decisions.
    DOI:  https://doi.org/10.1016/j.ijmedinf.2026.106561
  5. J Med Internet Res. 2026 Jul 02. 28 e92132
      The value of a meta-analysis is based on its methodological and statistical rigor, yet many published systematic reviews and meta-analyses contain statistical shortcomings that limit their utility for clinical practice and public health. This can make it challenging to aggregate data for treatment choices for patients as well as limit the extent to which policymakers can promote social change and improve public health. This challenge is compounded by the traditionally slow and resource-intensive nature of systematic reviews, which delays the translation of vital evidence. In this tutorial, we address both challenges. We first provide a primer on essential statistical techniques to help authors produce more robust and reliable meta-analyses. We then briefly discuss the growing role of artificial intelligence (AI) in automating tasks in systematic literature reviews and meta-analyses. Ethical use and disclosure of AI in supporting these essential tasks are also important considerations. This guide is intended to help authors enhance the rigor of their work and use new technologies to ensure their findings are both trustworthy and timely.
    Keywords:  artificial intelligence; evidence synthesis; heterogeneity; large language models; literature review; meta-analysis; systematic review
    DOI:  https://doi.org/10.2196/92132
  6. Environ Evid. 2026 Jul 02. pii: 9. [Epub ahead of print]15(1):
      We welcome Hodgson et al. (2026) empirical evaluation of ontology-grounded large language models (LLMs) for data extraction in environmental evidence synthesis. The study makes a valuable contribution by quantifying performance across attribute types and by openly documenting where current approaches struggle.
    DOI:  https://doi.org/10.1186/s13750-026-00388-7
  7. JMIR Med Inform. 2026 Jul 03. 14 e96129
       BACKGROUND: Qualitative thematic analysis is widely used in health research to examine patient experiences and inform the refinement of digital health interventions, but it is time- and labor-intensive. Large language models (LLMs) may help accelerate this process, yet their performance may depend not only on the model itself but also on how the analytic workflow is structured. Current evidence remains limited on how different LLMs perform across multistage thematic analysis workflows and across multiple health-related qualitative datasets.
    OBJECTIVE: This study aimed to evaluate a modular human-artificial intelligence (AI) collaboration pipeline for LLM-assisted thematic analysis and compare how model choice and workflow strategy influence alignment between AI-generated and human-generated themes across 3 qualitative health studies.
    METHODS: The framework was applied to analyze deidentified semistructured interview transcripts from 3 completed qualitative health studies involving patients with interstitial lung disease, postural orthostatic tachycardia syndrome, and chronic obstructive pulmonary disease. Three LLMs were compared: Gemini (Gemini 3 Pro), ChatGPT (GPT-5.2-thinking), and Opus (version 4.6). The workflow separated analysis into code extraction, code combination, and theme generation, and 5 strategies were tested. AI-generated themes were embedded using sentence-t5-xxl and compared with human-generated themes using cosine similarity after alignment with Hungarian and Greedy matching. Runtime and output-format consistency were also examined.
    RESULTS: Output volume differed substantially by model. Gemini generated the fewest codes and themes, while ChatGPT showed a similar but higher output ceiling. Opus produced the largest and most variable codebooks and theme sets. Across the 3 studies, Opus showed the strongest and most consistent alignment with human-generated themes, with the best cosine similarity scores observed in postural orthostatic tachycardia syndrome-direct coding (mean 0.893, SD 0.041), chronic obstructive pulmonary disease-direct grouping (mean 0.891, SD 0.027), and interstitial lung disease-L3 (mean 0.889, SD 0.032). ChatGPT was competitive in selected settings, whereas Gemini generally produced slightly lower similarity scores but had the shortest runtime. ChatGPT and Opus also showed better formatting consistency and workflow usability than Gemini.
    CONCLUSIONS: A modular human-AI pipeline can support thematic analysis across multiple digital health interview studies, but performance depends strongly on both model choice and workflow design. Opus produced the most consistently human-aligned themes, while Gemini and ChatGPT showed different trade-offs in speed, fidelity, and usability. These findings support the use of LLMs as structured, human-supervised analytic assistants rather than replacements for qualitative researchers.
    Keywords:  LLM; generative AI; health informatics; human-AI collaboration; large language model; qualitative study; thematic analysis
    DOI:  https://doi.org/10.2196/96129
  8. Front Res Metr Anal. 2026 ;11 1863790
      The qualitative research field is evolving as artificial intelligence (AI) and its applications gain broader attention. While these new technologies may be of help to some aspects of the qualitative research process, they also pose several challenges that must be reflected upon, addressed, and documented. Some of the techniques offered by AI for qualitative analysis have existed for a long time, but hallucinated outputs and the more nuanced predictive abilities that may violate privacy and ethical aspects are of major concern. As this is an emerging field, there is a need for a synthesis of recommendations on AI use for qualitative work to reduce risks to rigor and reproducibility. In this paper, we identify key ethical and practical concerns and provide recommendations across five key stages of the qualitative research process so researchers and stakeholders can safely leverage AI to complement their human expertise.
    Keywords:  artificial intelligence; critical reflections; ethical considerations; methodological challenges; mitigation strategies; qualitative analysis; qualitative research
    DOI:  https://doi.org/10.3389/frma.2026.1863790
  9. Res Integr Peer Rev. 2026 Jul 01.
       BACKGROUND: Statistical review is essential for research quality and integrity, yet traditional manual review is inefficient. Large language models (LLMs) offer potential support but are unreliable when used without guidance for precise calculations and raise concerns about accountability. This study evaluated whether a structured, rule-based prompt can reliably constrain an LLM to perform statistical review of comparative categorical data, and characterized both its feasibility and its inherent risks from an accountability perspective.
    METHODS: This study employed a two-stage design based on the DeepSeekV3.2. In the first stage, a structured prompt was developed through dozens of "test-fail-iterate" cycles using 20 published medical articles. The prompt assigned the LLM the role of a "statistics expert" and provided a closed set of computational rules and a "recognize data-select calculation formula-calculate" workflow for analyzing categorical data, including Pearson's Chi-square test, continuity correction, and McNemar's tests. In the second stage, the performance of the final prompt was evaluated on a test set of 20 independent manuscripts. The model's output was compared against the results calculated by a senior statistician (the gold standard). The primary outcome measures were the performance in statistical method selection and numerical computation, including accuracy, sensitivity (recall), specificity, positive predictive value, negative predictive value, F1 score, and Cohen's Kappa. Secondary measures included reproducibility and efficiency.
    RESULTS: The test set consisted of 15 manuscripts with independent samples and 5 with paired samples. In the assessment of the appropriateness of statistical method selection for 148 analysis items, the model achieved an accuracy of 99.3% (147/148), a sensitivity of 96.2% (25/26) (F1=98.0%, κ=0.976). For the test of computational consistency in 97 independent sample tests, the accuracy for χ2 value consistency was 94.8% (92/97) (F1=89.3%, κ=0.859), and for P-value consistency, it was 96.9% (94/97) (F1=90.9%, κ=0.891). In the paired-sample analysis, the model's methods and results were in perfect agreement with the manual review, and prompt optimization eliminated discrepancies in degrees-of-freedom calculation rules. Efficiency analysis showed no statistically significant difference in time consumption between the model (407 s) and manual review (374 s) (P=0.601). In reproducibility tests, the intraclass correlation coefficients for both χ2 values and P-values exceeded 0.91. However, qualitative analysis revealed 3 typical failure modes in the task workflow: (1) Instability: The model's failure to produce identical outputs across repeated runs, manifesting as inconsistent data extraction or the failure to process all designated tasks (scope neglect). (2) Performance degradation/"lazy" behavior: A decline in execution quality on long or complex tasks, often characterized by the model abandoning its reasoning process to copy author-provided values without verification. (3) Anchoring effect: The model's tendency to over-rely on author-provided statistical values (the "anchor"), causing its verification process to be unduly influenced.
    CONCLUSIONS: A structured, rule-based prompt can guide the DeepSeek to achieve high accuracy in standardized statistical review tasks, but its reliability is contingent on operational stability. Inherent failure modes, including performance instability and a strong anchoring effect on author-provided data, persist and can lead to significant errors, particularly when source data are flawed. These findings suggest that the the DeepSeek is not suitable for autonomous auditing. Their most appropriate application is as assistive tools within a human-in-the-loop framework, where rigorous human supervision is essential for risk mitigation and to maintain ultimate accountability.
    Keywords:  Accountability; Human-Computer Collaboration; Large Language Model; Prompt Engineering; Research Integrity; Statistical Review; Structured Prompt
    DOI:  https://doi.org/10.1186/s41073-026-00231-0
  10. Indian J Ophthalmol. 2026 07 01. 74(7): 1073-1076
       PURPOSE: To evaluate the accuracy and reliability of four artificial intelligence (AI) models-ChatGPT, Copilot, DeepSeek, and Gemini-in generating PubMed citations for literature related to lens disease, cataracts, iris disorders, and anterior chamber pathology.
    DESIGN: Comparative accuracy assessment study.
    METHODS: Forty standardized clinical paragraphs from The Review of Ophthalmology (4 th edition) were used as test inputs. Each AI model was prompted to generate AMA-11-style PubMed references. Citation accuracy was assessed using predefined criteria, including PubMed verifiability, DOI concordance, and bibliographic accuracy. Two expert reviewers independently classified the citations as fully cited, partially cited, or not cited, and assessed inter-rater reliability.
    RESULTS: The citation accuracy varied significantly among the models. DeepSeek demonstrated the highest accuracy (52.5%), followed by ChatGPT (32.5%) and Copilot (20.0%), whereas Gemini demonstrated the lowest accuracy (2.5%) ( P < 0.001). DOI mismatches were the most common errors across all models. Expert validation confirmed these findings, with DeepSeek producing the highest number of fully cited references. Inter-rater agreement was substantial (Cohen's κ = 0.65).
    CONCLUSION: Domain-specific AI models, particularly DeepSeek, outperform general-purpose models in generating PubMed citations from ophthalmic literature. However, all the evaluated models exhibited citation errors, underscoring the necessity of human verification. AI tools may enhance academic workflows as assistive systems but should not be used autonomously for reference generation in medical research.
    Keywords:  AI hallucinations; AI models; ChatGPT; Copilot; DeepSeek; Gemini; PubMed citations; anterior segment; artificial intelligence; cataract; citation generation; iris disorders; medical literature; ophthalmology; reference management
    DOI:  https://doi.org/10.4103/IJO.IJO_3155_25
  11. Stud Health Technol Inform. 2026 Jun 29. 338 242-246
      Large language models (LLMs) are increasingly explored for qualitative analysis, but the effect of workflow design on thematic fidelity remains unclear. This study evaluated a structured human-AI collaboration framework using Claude Opus 4.6 to analyze 16 interview transcripts from patients with chronic obstructive pulmonary disease participating in a pulmonary telerehabilitation program. The workflow included code extraction, code combination, and theme generation, and was tested using hierarchical and direct strategies. AI-generated themes were compared with human-derived themes using sentence-t5-xxl embeddings and cosine similarity, with theme alignment performed using Hungarian and greedy matching. Output volume varied substantially across strategies, ranging from 53 to 357 codes and 11 to 17 themes. Direct grouping (average cosine similarity 0.891) and L3 grouping (0.890) achieved the highest similarity to human-generated themes. These findings suggest that grouping-based workflows can preserve key information, reduce redundancy, and improve thematic generation in LLM-assisted qualitative analysis.
    Keywords:  Generative AI; LLM; Qualitative; Thematic Analysis
    DOI:  https://doi.org/10.3233/SHTI260838
  12. ArXiv. 2026 Apr 16. pii: arXiv:2604.15456v1. [Epub ahead of print]
      Trustworthiness and transparency are essential for the clinical adoption of artificial intelligence (AI) in healthcare and biomedical research. Recent deep research systems aim to accelerate evidence-grounded scientific discovery by integrating AI agents with multi-hop information retrieval, reasoning, and synthesis. However, most existing systems lack explicit and inspectable criteria for evidence appraisal, creating a risk of compounding errors and making it difficult for researchers and clinicians to assess the reliability of their outputs. In parallel, current benchmarking approaches rarely evaluate performance on complex, real-world medical questions. Here, we introduce DeepER-Med, a Deep Evidence-based Research framework for Medicine with an agentic AI system. DeepER-Med frames deep medical research as an explicit and inspectable workflow of evidence-based generation, consisting of three modules: research planning, agentic collaboration, and evidence synthesis. To support realistic evaluation, we also present DeepER-MedQA, an evidence-grounded dataset comprising 100 expert-level research questions derived from authentic medical research scenarios and curated by a multidisciplinary panel of 11 biomedical experts. Expert manual evaluation demonstrates that DeepER-Med consistently outperforms widely used production-grade platforms across multiple criteria, including the generation of novel scientific insights. We further demonstrate the practical utility of DeepER-Med through eight real-world clinical cases. Human clinician assessment indicates that DeepER-Med's conclusions align with clinical recommendations in seven cases, highlighting its potential for medical research and decision support.
  13. J Med Internet Res. 2026 Jun 29. 28 e88677
       BACKGROUND: Qualitative health research often focuses on how patients experience and manage chronic illnesses, a topic that has been extensively studied in the literature. With the emergence of large language models (LLMs), such as Claude (Anthropic PBC) and ChatGPT (OpenAI), new opportunities are arising to support and scale the thematic analysis of narrative health data. However, their role and added value compared to traditional human-led approaches remain underexplored, particularly in complex clinical contexts such as multimorbidity.
    OBJECTIVE: We aim to evaluate the methodological contribution of LLM-assisted analysis by examining its ability to replicate and extend established qualitative insights, in comparison with traditional thematic analysis.
    METHODS: Semistructured interviews were conducted with 30 individuals living with two or more chronic illnesses. Transcripts were analyzed using both manual thematic coding and Claude 3.5 Sonnet. A structured comparison was conducted to identify shared and unique themes across the two approaches. The analysis examined thematic overlap, differences in subtheme identification, and variation in the level of detail between the methods.
    RESULTS: Both approaches identified similar core themes related to the patient experience, including health care navigation and challenges, support systems and family dynamics, and emotional challenges and coping. Manual analysis produced more contextually detailed interpretations, while the LLM approach identified a larger number of subthemes. Each method also revealed distinct themes: the manual analysis included themes such as faith, caregiving roles, and a proactive mindset, whereas the LLM identified themes such as future planning and multiple health conditions. The findings show both similarities and differences between the two approaches. The LLM analysis also demonstrated efficiency in processing large volumes of qualitative data.
    CONCLUSIONS: A hybrid approach that integrates artificial intelligence-assisted and human-led thematic analysis can enhance both analytical depth and scalability. These findings support the use of LLMs as a complementary tool in qualitative research, while highlighting the importance of combining automated pattern detection with human interpretation.
    Keywords:  chronic illness; disease management; large language model; patient experience; qualitative research
    DOI:  https://doi.org/10.2196/88677
  14. J Med Imaging (Bellingham). 2026 May;13(3): 030101
      This editorial revisits the enduring role of literature reviews in scientific training and communication, emphasizing that their value lies in expert synthesis rather than simple aggregation of sources. In an era of AI-assisted discovery and summarization, the authors argue that peer-reviewed reviews must provide deeper perspective on how fields evolve and how evidence interconnects. They outline updated expectations for review and perspective articles, stressing that human judgment remains essential for interpreting and validating the scientific record.
    DOI:  https://doi.org/10.1117/1.JMI.13.3.030101
  15. Langenbecks Arch Surg. 2026 Jul 03. pii: 182. [Epub ahead of print]411(1):
       BACKGROUND: Amidst the current enthusiasm concerning artificial intelligence and its possible application in the composition of different kinds of scientific and non-scientific written documents, we evaluated the usage of artificial intelligence for writing surgical short reviews.
    METHODS: In order to assess the formal and content quality of AI-generated texts compared to human written texts, ten AI-based text generators (five chatbots and five content creators) and four surgeons in training received the same prompt for a short scientific article on a liver surgery theme. All texts were anonymized and subsequently evaluated by three experienced liver surgeons based on a pre-defined scoring scheme, as well as for quality of references and readability according to readability indices. Furthermore, all texts were tested for plagiarism using PlagScan.
    RESULTS: Overall percentage of correct assessment for AI/non-AI generation by experienced surgeons lay at 78.57%. Human written text had a mean word count of 1054 versus 874 in AI-generated text, with a higher mean Flesh Reading Ease Score (FRE, 26.2 ± 5.1 versus 17.7 ± 6.1). References were PubMed-listed in 100% for human versus 46% for AI-generated text, with only one non-human text reaching 100% formally correct citation of references. PlagScan found 6.4%±1.3 mean resemblance to existing texts for human versus 7.6%±4.5 for AI-generated text.
    DISCUSSION: Overall, AI could already mislead experienced scientific surgeons in 26.7% of cases into believing it to be human. However, formal requirements, especially considering referencing, are still in great need of improvement with only one of AI-generated articles fulfilling our quality requirements.
    Keywords:  AI-text generation; Hepatobiliary surgery; Large language model; Readability; Scientific writing
    DOI:  https://doi.org/10.1007/s00423-026-04095-2
  16. medRxiv. 2026 Jun 15. pii: 2026.06.06.26354746. [Epub ahead of print]
      Randomized controlled trials (RCTs) play a central role in assessing the benefits and harms of interventions. Incomplete reporting in RCT publications can compromise the verifiability and usefulness of RCTs. SPIRIT and CONSORT reporting guidelines aim to improve the completeness of RCT protocols and results publications, respectively. However, many RCTs are not reported completely. Checking manuscripts automatically could help authors improve the completeness of reports prior to publication. We previously annotated SPIRIT-CONSORT-TM, a corpus of 200 articles (comprising 100 protocol-results publication pairs) using 83 checklist items drawn from SPIRIT 2013 and CONSORT 2010. We also trained machine learning models to automatically assess reporting at the item level. Each checklist item can include multiple constituent elements (i.e., specific details required for that item), and an item might be considered fully reported when all of its elements are present. However, prior work does not explicitly capture or evaluate reporting at the element level. To address this gap, we extended SPIRIT-CONSORT-TM by incorporating element-level annotations and using them to assess reporting completeness (SPIRIT-CONSORT-ELM). We formulated element-level assessment as a machine reading comprehension task, operationalized through 119 questions, where each question targets a specific reporting element within a checklist item. Using the 200 articles included in SPIRIT-CONSORT-TM, two annotators independently answered 119 questions for 50 articles (25 protocol-results pairs) and resolved any discrepancies through discussion; the remaining 150 articles (75 protocol-results pairs) were assessed by a single annotator. We then developed an automated pipeline for element-level assessment using SPIRIT-CONSORT-ELM. The pipeline first applies a PubMedBERT-based model to identify sentences containing item-level reporting information, then it uses a generative large language model (LLM; GPT-5) with chain-of-thought reasoning to answer element-level questions based on the retrieved evidence. Agreement between the two annotators was high (Gwet's AC1: 0.782) and our pipeline achieved high accuracy in identifying element-level reporting evidence (F1: 0.822, Gwet's AC1: 0.796). Ablation studies indicate that chain-of-thought reasoning and the inclusion of illustrative in-context examples modestly improve LLM performance on the machine reading comprehension task. SPIRIT-CONSORT-ELM provides a benchmark for evaluating reporting guideline completeness at the element level, enabling assessment of RCT transparency beyond the simple presence or absence of checklist items and is publicly available at https://osf.io/kznx4/ . The automated pipeline establishes a robust baseline for assessing RCT reporting and demonstrates potential as a practical aid for authors, reviewers, and editors to identify and address gaps in completeness and transparency of RCT reports.
    DOI:  https://doi.org/10.64898/2026.06.06.26354746
  17. JAMIA Open. 2026 Jun;9(3): ooag082
       Objectives: To compare the performance of task-specific generative AI, with general-purpose large language models (LLMs) in generating more readable lay abstracts and summaries (LASs) of Oncology research.
    Materials and Methods: Twenty-five randomly selected abstracts from the top 5 journals in Oncology were processed into LASs using a task specific LLM-powered tool (Pub2Post) and 5 general-purpose LLMs (ChatGPT-5, Claude, Gemini, DeepSeek and Grok). Two prompting strategies (Specific and Generic) were applied. Consistency was tested across 3 outputs, producing a total of 825 LASs. Readability-scores and text-metrics were calculated. The "best test" per model was selected based on lowest SMOG Index, which was subsequently used to compare the 6 GAI platforms. Comparisons were performed using Kruskal-Wallis tests, with significance set at P < .05.
    Results: All the platforms demonstrated consistent intra-model outputs across triplicate generations (all P > .05). However, inter-model comparisons revealed significant differences (all P < .001) with Pub2Post outperforming the LLMs, across the 2 prompting styles, demonstrating superior readability-scores (FRES 82.3; FKGL 5.2; GFS 6.6; SI 4.4; CLI 10.4; ARI 6.2) with longer outputs (27 sentences; 388 words) and fewer complex-words (3.7%). The general-purpose LLMs generated shorter, denser outputs (4-9 sentences; 81-156 words) with higher grade-levels (FRES 38.0-61.6; FKGL 9.6-13.4; GFS 10.6-15.9; SI 7.6-11.2; CLI 13.4-17.0; ARI 11.0-15.2).
    Discussion and Conclusion: Task-specific GAI powered tools (Pub2Post) generated consistently more readable LASs compared to 5 commercially available LLMs, regardless of prompting strategy. These findings highlight the value of purpose-built GAI tools for enhancing public understanding and accessibility of oncology research, with implications for improving patient-education.
    Keywords:  GPT; GPT-5; Pub2Post; artificial intelligence; generative AI; large language models; lay summary; readability
    DOI:  https://doi.org/10.1093/jamiaopen/ooag082
  18. Cent European J Urol. 2026 ;79(2): 152-160
       Introduction: The aim of the study was to evaluate the accuracy of responses from three common artificial intelligence (AI) tools - ChatGPT, Claude, and DeepSeek - to patient enquiries regarding lower urinary tract symptoms (LUTS), comparing these responses to European Association of Urology (EAU) 2025 guidelines. As patients increasingly turn to large language models (LLMs), it is crucial to assess their reliability.
    Material and methods: Prospective face-to-face and telephone general urology clinics were conducted between April and May 2025, during which consented patients were asked to state the questions they would submit to an AI tool if they were to enquire about their LUTS. These questions were submitted to ChatGPT (GPT-4), Claude (Sonnet 4.0), and DeepSeek (V3), and the responses were summarised and compared against the EAU guidelines. Each response was categorised as correct, missing key elements from the guidelines, or incorrect/misleading.
    Results: Sixteen patients participated in the study, and following removal of duplicate questions, a total of 13 were included for analysis. These questions covered symptom causation, diagnostic workup and management of LUTS. All models provided correct information in 92.3% of their responses when compared to the EAU guidelines. However, 92.3-100% of answers omitted key elements from the guidelines, and 30.8-92.3% contained incorrect or misleading content.
    Conclusions: Current LLMs provide readily accessible guidance on LUTS; however, their unsupervised use in clinical decision-making remains a concern and may be considered premature.
    Keywords:  ChatGPT; Claude; DeepSeek; artificial intelligence; lower urinary tract symptoms
    DOI:  https://doi.org/10.5173/ceju.2026.0011
  19. bioRxiv. 2026 Jun 24. pii: 2026.06.19.732660. [Epub ahead of print]
      Biological research often requires information about species' traits. Manual literature collation can be time-consuming and miss parts of the literature. To address this gap, we developed trAIt , a publicly available software for the retrieval of characteristics of species from scientific literature catalogued in the Europe PubMed Central (PubMed) database. trAIt provides a graphical user interface (GUI) in which users specify species and characteristics of interest. Leveraging a large language model (LLM), trAIt retrieves relevant papers, combines their content through a consensus-based summarization model, and outputs a species-by-characteristic table. For a case study involving frog species, trAIt recovered 47.1% of trait-species combinations in 2.75 hours, while an expert curator independently recovered 62.4% over months. The consensus-based summarization substantially aids accuracy compared to single-source extraction. Across three case studies of vertebrate taxa, an expert confirmed the accuracy of 70.9% of trait-species entries recovered by trAIt . We observed considerable variation across taxa in trAIt 's accuracy, which is possibly due to heterogeneity in open-access literature availability and inconsistencies in species and trait terminology. In sum, our analysis suggests that LLM-based tools can accelerate biological data synthesis but should be used to support domain experts' research, rather than replace their judgment.
    DOI:  https://doi.org/10.64898/2026.06.19.732660
  20. BMC Res Notes. 2026 Jun 27.
       OBJECTIVE: A major problem with reviewing the statistical methodology in published medical articles is that extracting the necessary details from large sample sets is time-consuming. This paper demonstrates how a novel automated procedure can extract information about statistical reporting from literature. To illustrate this, we searched the PubMed Central database for original research articles published in 2021 and 2023 to identify the statistical software packages used for data analysis. A key element in terms of transparency and reproducibility is the reporting of the software used for statistical analysis.
    RESULTS: A freely available Shiny App was created with the help of generative artificial intelligence, and it was used to retrieve automatically information from randomly selected samples of articles indexed in PubMed Central. We analyzed a large sample of articles (n = 1740) to determine the reporting of statistical software for nine study designs. We found that, across different study types, proprietary software such as IBM SPSS Statistics still dominates. Despite multiple calls for greater use of open-source research software, these programs are not used as frequently. In addition, a surprising number of articles did not report the software used. Furthermore, this is the first application of the recent Vibe Coding concept to statistical research methods.
    Keywords:  Biostatistics; Generative artificial intelligence; Health Sciences; Medicine; Publications; Statistical reporting; Statistical software; Study designs
    DOI:  https://doi.org/10.1186/s13104-026-07908-1