Citations

BACKGROUND: Hypertension, the leading cause of cardiovascular mortality, remains suboptimally controlled. Large language models (LLMs) could improve hypertension control by augmenting clinical decision-making, but their reliability for guideline-driven tasks is unverified. This study evaluated the accuracy and safety of hypertension management recommendations generated by 3 LLMs. METHODS: Fifty-one vignettes were constructed and submitted to the LLMs (GPT-4, Gemini, Medical Large Language Model [by Google; MedLM]) and a hypertension expert to generate the responses. Three blinded reviewers rated each response on a 4-point accuracy scale, a binary safety (safe/unsafe) scale, and attempted to identify the source (LLM versus expert) providing the response. RESULTS: GPT-4 had the highest accuracy (83%) and safety (86%) scores among LLMs but remained inferior to expert responses (92% accuracy, 93% safety). Gemini and MedLM performed significantly worse (accuracy: 64% and 35%; safety: 73% and 39%, respectively). GPT-4 generated the most guideline-concordant responses (46%) among the 3 LLMs (Gemini 35%, MedLM 14%) but was lower than expert responses (68%). Interrater reliability for accuracy ratings was higher for LLM-generated responses (GPT-4 [intraclass correlation coefficient, 0.30], Gemini [intraclass correlation coefficient, 0.61], and MedLM [intraclass correlation coefficient, 0.58]), with lower agreement for expert responses (intraclass correlation coefficient, 0.23). A similar pattern was observed for safety and source discrimination ratings. The agreement was strongest for safety assessments and weakest for source discrimination. CONCLUSIONS: Among the 3 tested LLMs, GPT-4 demonstrated closer agreement to expert decisions, thereby showing greater potential for supporting hypertension management. Despite their potential, current LLM versions are inferior to expert recommendations. Human-in-the-loop supervision remains essential when deploying LLMs for clinical decision-making.

Abstract Objective: The National Institutes of Health (NIH) Executive Abilities: Measures and Instruments for Neurobehavioral Evaluation and Research (EXAMINER) is a validated laptop-based battery of executive functioning tests. A modified tablet version of the EXAMINER was developed on the UCSF Tablet-based Cognitive Assessment Tool (TabCAT-EXAMINER). Here we describe the battery and investigate the reliability and validity of a composite score. Methods: A diagnostically heterogeneous sample of 2135 individuals (mean age = 65.58, SD = 16.07), including controls and participants with a variety of neurodegenerative syndromes, completed the TabCAT-EXAMINER. A composite score was developed using confirmatory factor analysis and item response theory. Validity was evaluated via linear regressions that tested associations with neuropsychological tests, demographics, clinical diagnosis, and disease severity. Replicability of cross-sectional results was tested in a separate sample of participants ( n = 342) recruited from a frontotemporal dementia study. As this separate sample also collected longitudinal TabCAT-EXAMINER measures, we additionally assessed test-retest reliability and associations between baseline disease severity and changes in TabCAT-EXAMINER scores. Results: The TabCAT-EXAMINER score was normally distributed, demonstrated high test-retest reliability, and was associated in the expected directions with independent tests of executive functioning, demographics, disease severity, and diagnosis. Greater baseline disease severity was associated with more rapid longitudinal TabCAT-EXAMINER decline. Conclusions: The TabCAT-EXAMINER is a tablet-based executive functioning battery developed for observational research and clinical trials. Performance can be summarized as a single composite score, and results of this study support its reliability and validity in cognitive aging and neurodegenerative disease cohorts.

Background Outcome selection in regional anesthesia and acute pain research is inconsistent, often lacking patient-centered priorities and validated instruments. We aimed to prioritize key outcomes and propose suitable measurement tools to improve the quality, consistency, and relevance of regional anesthesia research. Methods We conducted a multiround Delphi process, including two electronic voting rounds, a multiday in-person meeting, and a patient panel. Experts and patients evaluated existing and proposed outcomes for importance, validity, and relevance. Outcomes reaching ≥70% agreement were prioritized, and appropriate measurement instruments were identified. Results Thirty-two experts and three patients participated. Across three Delphi rounds, 10 outcomes were prioritized for future regional anesthesia research: seven efficacy outcomes (pain scores; opioid consumption; Brief Pain Inventory; functional outcomes; cognitive function; length of stay; block duration); two multidimensional outcomes (quality of recovery and development of a regional anesthesia-specific quality-of-recovery instrument); one safety outcome (chronic postsurgical pain). Patients prioritized cognitive recovery, function, chronic pain, and pain scores. Conclusions This consensus-based, patient-centered framework defines the core priorities for future regional anesthesia and acute pain research. Future work should focus on developing and validating a regional anesthesia-specific quality-of-recovery instrument, refining multidimensional measures such as functional recovery, cognitive outcomes, and the Brief Pain Inventory, and standardizing the reporting of chronic pain and opioid-related outcomes. Adoption of these priorities will enhance methodological consistency, comparability, and patient relevance in future clinical trials.

Introduction Implementation science research increases the uptake of evidence-based interventions, which may improve health equity among racial and ethnic minorities. However, it is unclear how anti-racism and anti-colonialism practices have been integrated into implementation science research. The objectives of this scoping review are to describe the current conceptualisations of racism and colonialism within the USA, examine racism or colonialism-conscious approaches and analyse gaps in the operationalisation of anti-racism or anti-colonialism within implementation science studies. Methods and analysis This scoping review will be conducted following the Preferred Reporting Items for Systematic Reviews and Meta-analyses extension for scoping reviews guidelines. The Center for Chronic Disease Reduction and Equity Promotion Across Minnesota conceptual framework and an implementation science anti-racism lens will guide the study design and analysis. To determine study eligibility for the scoping review, articles will undergo abstract and full-text screening by two independent reviewers and discrepancies will be settled together. Data charting will be extracted from included articles by eight independent reviewers. The search strategy will use controlled vocabulary and natural language keywords related to health equity, health disparities and anti-racism/colonialism on six databases. The scoping review will include studies that applied implementation science theories, models or frameworks among US-based populations. Additionally, included studies will report any of the following implementation activities: implementation strategies, implementation outcomes, adaptations to evidence-based interventions, or evaluations of pre-implementation or implementation context. Ethics and dissemination No ethical approval was required for the scoping review. Dissemination will be through publication in a peer-reviewed journal and conference presentations.

Background: The PRO-CTCAE provides patient-reported data on symptomatic AEs. A summary metric—the ACS—reflecting total AE burden can be calculated by averaging AE-level composite scores at a given timepoint for each participant. This study investigated the psychometric properties and interpretability of this PRO-CTCAE ACS in patients with breast, lung, or head/neck cancers. Methods: We conducted a secondary analysis of a PRO-CTCAE validation dataset comprising 940 adults undergoing chemotherapy or radiation therapy (clinicaltrials.gov: NCT02158637). We focused on empirically recommended symptom terms for three cancer sites. Analyses included Spearman’s correlations, coefficient alpha, and eigenvalues from the correlation matrices, confirmatory factor analysis (CFA), and principal component analysis (PCA). Latent profile analysis (LPA) was used to assess ACS interpretability in the lung cohort. Results: Mean composite score inter-correlations were moderate (0.30–0.35), and coefficient alphas were high (0.81–0.91). Eigenvalue ratios and CFA supported retention of a single factor/component, with suitable model fit indices. ACS correlated highly with factor scores and the first principal component from the PCA. Reduced sets of terms produced reliable scores that closely approximated the full set scores and aligned with external criteria. LPA in the lung subgroup identified four latent classes; ACS differentiated high vs. low symptom burden groups but did not distinguish the two groups expressing distinct symptom profiles. Conclusion: The ACS demonstrated structural validity through adequately fitting linear factor models and effectively summarized symptomatic AE burden. However, similar ACS values may mask clinically distinct symptomatic AE profiles, underscoring the value of both summary metrics and profile-based approaches.

ABSTRACT Suicidality contributes substantially to global morbidity and mortality, yet despite its heritability, its biological etiology remains largely elusive. We conducted multi-ancestry genome-wide association study meta-analyses of suicidal ideation (259,747 cases), suicide attempt (64,993 cases), suicide death (9,197 cases), and suicidal behavior (suicide attempt/ death, 75,300 cases), across 54 cohorts (e.g., the Psychiatric Genomics Consortium, Million Veteran Program, UK Biobank). We identified 77 significant loci across meta-analyses, including 59 previously unreported for suicidality. SNP-based heritability ranged from 2.0-6.7% and there were strong, yet incomplete, genetic correlations between suicidality phenotypes (0.70-0.88). Fine-mapping prioritized putative causal SNPs and 20 credible genes. Enrichment analyses implicated synaptic pathways and neuronal populations predominantly in subcortical brain regions (e.g., amygdala excitatory, medium spiny, hippocampal CA1-3). Together, these findings establish suicidality as a polygenic set of traits with both shared and distinct genetic influences, providing a foundation for future studies of suicide biology and etiology.

Abstract Disclosure: E.M. Lewiecki: Amgen Inc, Kyowa Kirin, Ultragenyx, Angitia, Radius Health, Inc, Ascendis. E. Kapoor: Fresenius Kabi USA Inc., Astellas Pharma, Estetra SRL, WellFound Inc., Exeltis. S. Hurtado: None. V.M. Navarro: Bayer, Inc. M. Torres: Bayer, Inc. J. Beau: Bayer, Inc. I. Gkioni: Syneos Health LLC (providing service to Bayer), AstraZeneca. K. Genga: Bayer, Inc.. Background: The neuropeptide substance-P and its receptor neurokinin-1 (SP-NK1R pathway) play a role in bone metabolism and remodelling. Pre-clinical evidence also points to a role of this pathway in energy homeostasis. NK-1R antagonists have been proposed as potential therapies in obesity and musculoskeletal disorders like osteoporosis. Elinzanetant (EZN) is a dual neurokinin-1 and -3 receptor antagonist in late-stage development for vasomotor symptoms (hot flashes) associated with menopause. This post hoc analysis evaluates potential benefits of EZN on the skeleton, body weight, and body composition in postmenopausal women. Methods: Data were gathered from exploratory endpoints of the 52-week (wk), placebo (PL)-controlled, OASIS-3 Phase 3 EZN efficacy and safety trial. Mean percentage (%) change in total bone mass (kg) (measured indirectly by bioelectrical impedance analysis at baseline (BL), wks 4, 8, 12, 18, 24, 36, 52); mean % change in bone mineral density (BMD, measured directly by dual-energy X-ray absorptiometry at BL, wks 24 and 52) in femoral neck, total hip, and lumbar spine; and mean % change in key bone turnover markers osteocalcin (OC) and procollagen 1N-terminal propeptide (P1NP) at BL, wks 4, 8, 12, 18, 24, 36, 52 were evaluated. Weight, body mass index (BMI: weight/height2), waist circumference (WC), fat mass index (FMI; fat mass/height2), lean mass index (LMI; lean mass/height2) and body water were measured to monitor body composition at BL and wks 4, 8, 12, 18, 24, 36, 52. A sub-analysis was performed based on FMI at baseline (FMI ≤9 normal fat mass; >9 excess fat mass). Descriptive statistics were used to evaluate all parameters and trends. Results: From BL to wk 52, increasing trends in bone mass were observed with EZN (mean [SD]: 2.81 [1.02] to 2.95 [1.96]) but not PL (mean [SD]: 2.87 [1.09] to 2.87 [1.13]). In parallel, lower BMD reductions were seen in EZN vs PL at all skeletal sites [EZN vs PL, mean % change from BL: femoral neck (0.00% vs -1.22%), total hip (-0.70% vs -1.37%), and lumbar spine (-0.57% vs -1.22%), with considerable variability in both groups (SD range 3.01 - 4.70)]. Lower levels of circulating OC and P1NP were consistently observed over 52 wks of EZN treatment, suggesting a lower rate of bone resorption with EZN vs PL. Trending reductions in weight, BMI, and WC, along with increases in LMI and body water were observed with EZN vs PL. Trends of higher LMI in patients with excess fat, or lower FMI in patients with normal fat at BL were observed with EZN, suggestive of a positive body re-composition with EZN. Conclusions: These data expand on EZN safety profile and further point to novel potential benefits for the skeleton and body composition of postmenopausal women. Further research is warranted to provide conclusive evidence of the benefits of EZN specifically in post-menopausal women with obesity/overweight status or osteoporosis. Sponsorship: Bayer U.S., Whippany, NJ, USA Presentation: Monday, July 14, 2025

Acute severe ulcerative colitis (ASUC) is a medical emergency affecting up to 25% of patients with ulcerative colitis (UC), with colectomy required in approximately 25–30% of cases during the initial admission. Intravenous corticosteroids remain the first-line therapy, though one-third of patients do not respond, necessitating rescue with infliximab or calcineurin inhibitors, which are both supported by randomized trials and guideline recommendations. Comparative studies and meta-analyses have shown similar efficacy between these agents, while sequential use is associated with higher adverse event rates and should be restricted to specialized centers. Recent data have refined infliximab use, with the PREDICT-UC trial showing no superiority of intensified dosing over standard regimens. Emerging therapies are under investigation: vedolizumab has been used as maintenance following calcineurin induction; ustekinumab has shown benefits in retrospective UC cohorts, particularly after cyclosporine; and Janus kinase (JAK) inhibitors represent the most recent addition. The randomized TACOS trial and the prospective TRIUMPH study demonstrated an improved short-term response with tofacitinib in steroid-refractory ASUC, and real-world reports suggest promising outcomes with upadacitinib. While infliximab and cyclosporine remain as standard rescue therapies, ongoing trials with novel agents are likely to broaden treatment options. This review summarizes the clinical features, initial management, and the role of advanced therapies in ASUC.