Background: Generative artificial intelligence (GenAI) tools are increasingly used in scientific research to support literature searches, evidence synthesis, and manuscript preparation. While these systems promise substantial efficiency gains, concerns have emerged regarding their reliability, particularly their tendency to cite inaccurate, fabricated, or retracted literature. The unrecognized inclusion of retracted studies poses a serious risk to research integrity and evidence-based decision-making. Whether commonly used GenAI tools can reliably detect, exclude, or transparently communicate the retraction status of scientific publications remains unclear. Objective: This study aimed to evaluate the ability of freely available GenAI tools to correctly handle retracted scientific articles during literature searches. Primary and secondary outcomes focused on accuracy, reliability, and consistency in recognizing retracted literature. Methods: In this pragmatic trial, nine widely used free-access GenAI tools (ChatGPT 4, ChatGPT 5, Claude, Gemini, Perplexity, Microsoft Copilot, SciSpace, ScienceOS, and Consensus) were evaluated. Each tool was asked five predefined, standardized questions addressing topic overview, article identification, article summarization, and explicit assessment of retraction status. Overall, 15 retracted articles (the 10 most cited and 5 most recently retracted as of May 23, 2025) were selected from the Retraction Watch database. All questions were repeated twice to assess intratool consistency. Responses were independently rated as correct or incorrect by 2 researchers. Descriptive statistics summarized performance, and comparisons between general-purpose and research-focused AI tools were conducted using descriptive statistics. Interreviewer agreement was assessed using Cohen kappa coefficient. Results: None of the evaluated AI tools consistently handled retracted articles correctly. No model achieved perfect accuracy across all question sets. ChatGPT 5 performed best, defined by the primary outcome of achieving fully correct responses to all five predefined tasks (5/5) for the highest number of retracted articles, correctly answering all five questions for 8 of 15 articles (53.3%). Research-focused tools (SciSpace, ScienceOS, and Consensus) failed to produce a single fully correct response set. Retracted articles were frequently included in topic overviews without warning, with error rates exceeding 40% in several tools. When specifically asked about retraction status, most systems failed to provide correct or complete information. OpenEvidence only reported data for a subset of our retracted articles as it is only used in health care literature. It demonstrated strong performance in topic overviews but low accuracy in identifying retracted articles. Conclusions: Freely available GenAI tools are currently not able to detect, exclude, or appropriately flag retracted scientific literature. The widespread and confident reproduction of retracted studies represents a substantial threat to research integrity, particularly in medical and evidence-based fields. Until retraction-aware verification mechanisms are systematically integrated, independent source checking remains essential when using AI-assisted literature tools.

High sound levels in intensive care units lead to increased rate of delirium and sleep disturbances in patients but also decreased work performance of healthcare providers. It is unclear how much impact different architectural settings have. The study aimed to determine whether sound levels in ICUs differ between single- and double-bed rooms. This prospective observational study collected sound level data in three architecturally identical ICUs at the Medical University of Graz. Recordings were made every 5 s, then aggregated into 1-minute and subsequently hourly intervals. For each hour, mean (LAeq), minimum (LAmin), and maximum (LAmax) sound pressure levels and standard deviations were calculated. Hourly sound levels between single- and double-bed rooms were compared by independent-samples t-tests. Secondary analyses examined day-night differences. All values were compared against WHO guideline thresholds. A total of 188 ICU patients were included, generating over 11 million sound data points. Contrary to expectations,single-bed rooms exhibited significantly higher LAeq(51.85 dBA vs. 50.29 dBA;p = 0.026),higher LAmin(41.22 dBA vs. 35.34 dBA;p < 0.001), andlower LAmaxcompared to double-bed rooms (67.80 dBA vs. 70.89 dBA;p = 0.023). Across all room types, sound levels were significantly higher during daytime for all parameters. None of the hourly values met WHO sound thresholds. Single-bed ICU rooms showed higher average and baseline sound exposure than double-bed rooms, challenging assumptions about acoustic advantages of single occupancy. Sound levels always remained above international recommendations. These findings highlight the need for multimodal strategies to reduce ICU sound, including staff awareness, alarm management, and architectural considerations. Structural design alone is insufficient; targeted behavioural and environmental interventions are essential to create quieter ICU environments.

Background: Boswellia serrata has traditionally been used in Ayurvedic medicine for its anti-inflammatory and antioxidant properties. Although several studies support clinical analgesic efficacy, the underlying mechanisms have not been investigated in human experimental pain models. This randomized, double-blind, placebo-controlled, crossover pilot trial aimed to examine the mode of action of Boswellia serrata to differentiate between its peripheral and central effects. This exploratory pilot study was designed to generate preliminary effect size estimates and assess functional pain-processing outcomes, rather than to provide definitive evidence of clinical efficacy. Methods: Twelve healthy volunteers were recruited and received either 300 mg of Boswellia serrata extract or a visually identical placebo twice daily for 28 days, separated by a 4-week washout period. Pain and sensitization were induced using a topical capsaicin model. Outcomes included spontaneous pain intensity, mechanical allodynia, pinprick hyperalgesia, thermal thresholds, and conditioned pain modulation, alongside psychological assessments of mood, anxiety, sleep, and structured adverse-event monitoring. Results: Results showed no significant difference in the primary endpoint of spontaneous pain intensity between Boswellia and placebo (VAS 43 ± 21 vs. 47 ± 17; d = 0.18; p = 0.539). Conclusions: While Boswellia serrata did not significantly reduce acute peak pain in this model, the observed trends suggest a potential multi-level modulatory influence on nociceptive processing and endogenous pain inhibition. These findings warrant larger clinical trials to further elucidate its therapeutic potential, particularly in populations with impaired pain modulation.

Background/Objectives: Accurate visual estimation of blood loss during childbirth is critical for early recognition of obstetric hemorrhage, a leading cause of maternal morbidity and mortality. Despite its widespread use, visual estimation is prone to substantial bias. While professional experience has been shown to influence estimation accuracy, the potential contribution of gender differences remains insufficiently explored. Methods: We carried out a prospective observational simulation-based study at a tertiary university medical center in Graz, Austria, to determine whether gender and professional background are associated with differences in the accuracy of visual blood-loss estimation in obstetric bleeding. Fifty physicians (28 females/22 males) were recruited from anesthesiology and obstetrics. All participants visually estimated blood loss in four simulated obstetric hemorrhage scenarios. The simulated blood volumes and hemoglobin concentrations were verified by volume measurement and point-of-care testing. Each scenario was viewed under standardized conditions without access to physiological or contextual clinical information. The primary outcome was absolute estimation error (mL) according to the gender or professional background of the participants. Secondary outcomes included scenario-specific accuracy and the association between self-rated confidence and estimation accuracy. Results: Women outperformed men in low and moderate-volume scenarios. Across all scenarios combined, women demonstrated lower median absolute estimation errors, although the overall difference did not reach statistical significance. Professional background showed a stronger effect: gynecologists were significantly more accurate than anesthetists across most scenarios. Conclusions: Visual blood-loss estimation accuracy in obstetric simulations is influenced by both gender and professional background. Gender-related differences appear volume-dependent, whereas professional experience exerts consistent influence.

Background and Importance: Early trauma‐induced coagulopathy (eTIC) is a major contributor to hemorrhage‐related mortality after severe trauma. Viscoelastic testing using thromboelastography (TEG) has been increasingly implemented in trauma resuscitation; however, its prognostic value compared with conventional coagulation testing remains uncertain. Objectives: We evaluated the ability of TEG parameters to detect eTIC, hypofibrinogenemia, and hyperfibrinolysis in severely injured trauma patients. In addition, we assessed their prognostic performance for transfusion requirements and mortality compared with conventional laboratory parameters. Design: Retrospective single‐center cohort study. Settings and Participants: Severely injured adult trauma patients admitted to the resuscitation room of a Level I trauma center, between January 2019 and October 2024 were screened. Outcome Measures and Analysis: Primary outcomes were detection of eTIC (INR > 1.2), severe coagulopathy (INR > 1.5), hypofibrinogenemia (< 200 mg/dL), and thrombocytopenia (< 100,000/μL). Secondary outcomes included the ability of coagulation parameters to predict the need for transfusion of ≥ 5 units of packed red blood cells within the first 24 h after admission and in‐hospital mortality. Results: A total of 262 patients were included; median ISS was 29 (IQR 23–41), and in‐hospital mortality was 22.8%. eTIC was present in 31.5% of patients. TEG parameters showed moderate discriminatory performance for coagulopathy detection. For eTIC, maximum amplitude achieved an AUC of 0.73, while R‐time achieved an AUC of 0.65. Predictive performance improved for severe coagulopathy, with R‐time reaching an AUC of 0.81. Hyperfibrinolysis (LY30 > 7%) was uncommon (6.9%). Conventional laboratory parameters, particularly INR, consistently outperformed TEG‐derived parameters for prediction of transfusion requirements (AUC 0.85) and mortality (AUC 0.79). Conclusion: In severely injured trauma patients, conventional coagulation testing, particularly INR, demonstrated superior prognostic performance compared with individual TEG parameters. Viscoelastic testing may provide complementary information but did not replace standard laboratory testing for prediction of transfusion needs or mortality in this cohort. Trial Registration: ClinicalTrials.gov identifier: NCT07186712