Hadassah Academic College

Large Language Models (LLMs) demonstrate increasing capabilities in medical knowledge assessment, yet limitations remain in cross-population validation, direct human-AI comparisons, and evaluation of newer models in anesthesiology contexts. This study addresses these gaps by conducting a head-to-head comparison between newer LLMs and human examinees on official Israeli multiple-choice board examinations. We evaluated two LLMs (Claude 3.7 Sonnet and ChatGPT-4) against anonymized aggregate data from 381 examinees on three consecutive official Israeli anesthesiology board examinations (2023–2024), comprising 450 multiple-choice questions stratified by difficulty, discrimination ability, and topic. Each model was tested twice per exam. Claude 3.7 Sonnet achieved 73.67% accuracy, significantly outperforming both human examinees (62.77%, P < 0.001) and ChatGPT-4 (64.44%, P < 0.001). However, both LLMs performed below the upper quartile of human performance (78.05%). While LLMs excelled on easy questions and theoretical domains like cardiac physiology (Claude: 96.88%, ChatGPT-4: 81.25%), they showed lower performance in areas such as ambulatory (Claude: 30.00%, ChatGPT-4: 10.00%) and regional anesthesia (Claude: 44.44%, ChatGPT-4: 38.89%). Human examinees demonstrated consistent performance across all domains, whereas LLMs showed extreme variability. Self-consistency was substantial for both LLMs (κ = 0.66–0.68), but agreement with human responses was moderate (κ = 0.34–0.39). While advanced LLMs currently exceed average examinee performance on anesthesiology board examinations, they fall short of top-quartile examinees at present and demonstrate significant performance variability across different topic areas.

Metabolomic studies in COPD reveal systemic metabolic perturbations, yet sex is often treated as a covariate rather than a biological driver. We aimed to identify plasma metabolites differentiating COPD from controls and to define sex-specific metabolic signatures in both groups. Methods: In this controlled observational study (BIOMEPOC cohort), untargeted plasma metabolomics was performed by LC-MS/MS. Differential abundance was tested across four contrasts (COPD vs. controls; men vs. women within controls; men vs. women within COPD; sex-by-disease interaction) with a false discovery rate (FDR) correction. Because smoking history differed between COPD and controls, a post hoc ever-smokers analysis was conducted. Results: COPD differed from controls in nine metabolites (all decreased): DL-stachydrine, 3-methyl-L-histidine, fructose, pipecolinic and nipecotic acids, 5-nitro-o-toluidine, conjugated linoleic acid, aminoadipate, and creatinine. This pattern is compatible with metabolic depletion, remodeling, and/or altered flux across multiple compartments rather than simple substrate deficiency, spanning muscle-related pools, amino acid handling, carbohydrate-associated metabolism, and exposome-linked inputs. In ever-smokers, results were directionally consistent, with five metabolites remaining nominally significant. Among controls, five metabolites were higher in men after FDR correction (PABA, cis-4-hydroxy-D-proline, N-acetylasparagine, deoxycarnitine, and creatinine), consistent with physiological sex dimorphism in energy pathways, connective-tissue remodeling, and diet/microbiome-related metabolism. Within COPD, six metabolites differed by sex after FDR correction, defining three axes: creatine energy buffering (men: higher GAA/creatinine, lower creatine), purine/urate handling (men: higher urate), and conjugated bile acids (men: higher GCDCA), implicating muscle bioenergetics, redox/inflammatory tone, and gut-liver crosstalk. Conclusions: Plasma metabolomics identifies a pattern compatible with systemic remodeling in COPD and sex-associated divergences in creatine, purine/urate, and bile-acid pathways, supporting a sex-influenced view of systemic COPD heterogeneity and highlighting targets for mechanistic validation.

As post-mortem (PM) imaging in children becomes more common, there is a need to review the available evidence for its diagnostic yield in suspected child physical abuse. The aim of this review is to synthesise current evidence, assess study quality, and identify ongoing challenges. Following PRISMA guidelines, databases were searched until 31 December 2024. Original research articles reporting data on at least ten children with PM imaging in the context of physical abuse were included. Titles and abstracts were screened by two expert reviewers; full texts were assessed by a third, independent reviewer and one of the previous reviewers. Data was extracted by one of 12 experts and independently verified. The study risk of bias was evaluated with the ROBINS-I tool. Study heterogeneity precluded meta-analysis, resulting in descriptive synthesis. Eighteen out of 1687 potential papers were included. Seven described PM radiography, five post-mortem computed tomography (PMCT), four both PM radiography plus PMCT, and two post-mortem magnetic resonance imaging (PMMR). All but one were retrospective, and most (11/18, 61%) had a moderate-to-high risk of bias. Post-mortem skeletal survey (PMSS) detected subtle fractures, particularly corner metaphyseal fractures. PMCT provided a high-resolution assessment of injuries, particularly rib fractures. PMMR contributed soft-tissue and intracranial detail. All studies emphasised the importance of correlating autopsy findings. Technical variation and potential biases limited direct comparisons between studies. PM imaging can reveal important injury patterns that may be overlooked by autopsy. Nevertheless, standardised imaging methods and larger prospective trials are needed to reduce bias and establish best-practice guidelines. Question What is the evidence for PM radiologic imaging in suspected physical abuse of children? Findings PM imaging complements autopsy, but diagnostic accuracy varies by modality. Study heterogeneity and bias limit current evidence. Clinical relevance PM imaging can detect injuries missed at autopsy in child abuse cases. Standardised protocols and higher-quality studies are urgently needed.

IntroductionSocial robots are increasingly explored as non-pharmacological support for older adults with Mild Cognitive Impairment (MCI) and dementia, yet their day-to-day integration remains limited. This study centers on designing an affective care model that empowers caregivers to deploy a commercial humanoid robot.MethodsAn exploratory qualitative study comprised four observation sessions and ten in-depth, open-ended interviews. A two-phase observational protocol was conducted in a day-care center: Phase 1 mapped everyday activities and baseline interaction patterns; Phase 2 documented two robot-assisted sessions. Detailed descriptions underwent iterative thematic analysis to extract themes and construct a care model grounded in caregivers' practices.ResultsIntroducing the robot added care practices that reshaped group dynamics. Two overarching themes emerged: (1) stimulation, engagement, and reciprocity and (2) shared experience and resonance. Together they embody a person-centered, affect-oriented approach that guides caregivers in meeting residents' emotional needs. These themes were synthesized into a structured three-step model, triggering, imagining, responding, that shows how interaction unfolds and how learning is refined through continuous feedback.DiscussionThe proposed care model sequence offers an actionable framework enabling caregivers to transform robotic interaction into a practical, intuitive tool. It provides a new method for training staff to integrate and adapt off-the-shelf robots within routine care, enhancing human-to-human engagement by capturing attention and stimulating memories. Given the global shortage of caregivers, empowering them to run structured robot-assisted sessions offers a scalable, cost-effective solution for health and care organizations working with people living with MCI and dementia.

The current study examined the neural mechanisms underlying auditory outcomes in pediatric cochlear implant (CI) users using event-related potentials, focusing on the frontocentral Novelty P3 (indexing automatic attention) and the parietal Target P3 (reflecting controlled attention). By comparing CI children with normal-hearing peers, we aimed to elucidate how these attention mechanisms contribute to post-implantation auditory performance variability. Prelingually deaf children with early bilateral CIs (n = 25) and normal-hearing (NH) controls (n = 28), matched for gender and age (7–13 years), completed an oddball paradigm during ERP recording and performed speech-in-noise and memory tasks. CI users showed reduced Novelty P3 amplitudes compared to NH peers, suggesting impaired automatic attention mechanisms, while Target P3 amplitudes were comparable between groups, indicating preserved controlled attention. CI users demonstrated poorer performance on speech-in-noise recognition and forward digit span tasks. Larger Novelty P3 and Target P3 amplitudes both correlated positively with speech-in-noise performance across all participants. In contrast, in CI users specifically, Target P3 amplitudes showed positive associations with short-term memory performance. This study provides novel evidence that pediatric CI users exhibit reduced automatic attention shifting in response to acoustic changes while maintaining intact controlled attention processes that may serve as a compensatory mechanism. The neural markers’ associations with behavioral performance suggest potential intervention targets: rehabilitation efforts might focus on strengthening automatic attention shifting while leveraging preserved controlled attention pathways.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-24476-9.

In a society whose needs are constantly changing, family planning plays a central role for women, men, and sustainable development. This comprehensive review and position statement summarises the proceedings of a meeting on contraception held in Rome in March 2024, supported by major scientific societies in the field. The aim is to inform the medical community about current medical and ethical issues of contraception use. First, the review addresses the complex ethical, religious, and social dimensions of contraceptive use and access; second, it provides a comprehensive analysis of traditional and modern contraceptive methods, discussing their safety and effectiveness; third, it examines current knowledge about male hormonal contraception. When prescribing a contraceptive method, medical indications or contraindications must be integrated to women’s religious beliefs, the geopolitical context in which they live, the risk of violence, their need for self-determination and their right to make decisions for themselves. If a partner is involved, the couple’s dynamics and shared needs must be considered. Healthcare providers are responsible for providing them with all the information they need to make informed choices, while ensuring individual autonomy. This position statement provides recommendations on how to guide contraceptive choice and identifies knowledge gaps about contraception today.

Objectives: Operative fixation of displaced midshaft clavicle fractures has become increasingly the more acceptable choice of care in recent years, based on evidence supporting its effectiveness. However, this practice presents challenges due to the complex S-shaped morphology of the clavicle and its subcutaneous location. Despite the introduction of anatomically pre-contoured plates, achieving optimal implant-to-bone fit remains difficult, prompting the development of newer plate generations. The aim of this study was to compare the hardware removal rates of second-generation 2.7 mm thinner plates (SGPs) with those of first-generation 3.5 mm plates (FGPs). Methods: A retrospective comparative cohort study was conducted at a level one trauma center. A total of 187 patients received FGPs, and 67 received SGPs, both positioned on the superior bone surface. All surgeries were performed by fellowship-trained surgeons, and patients were followed for at least one year. Data were extracted from medical records and the PACS system. Results: The patients’ demographics (age: 32.86 vs. 33.14 years; gender: 16.85% vs. 14.92% female) and fracture type (AO/OTA) were similar between the two groups. The rate of implant removal (20.1% vs. 20.9%) did not differ significantly between groups. Complications included nonunion (1.6% vs. 1.7%, statistically not significant) and infection (three cases in the FGP group, none in the SGP group). Conclusions: Despite the high success rate of clavicle fixation procedures, the superior clavicular plate removal rate remains significant, regardless of the implant design.

Abstract A consistent theory of quantum entanglement requires that the constituent single-particle states belong to the same Hilbert space, the coherent eigenstates of a complete set of operators in a given representation, defined with respect to a shared continuous parameterization. A covariant theory of relativistic spin and entanglement has been given by Horwitz and Arshansky, in which representations of O(3, 1) are induced with respect to an arbitrarily selected timelike unit vector nµ . In this paper, we construct an induced relativistic representation of spin on an extended phase space {(xµ , pµ ), (ζµ , πµ )}, in which we associate the unit vector nµ with the momentum πµ , thus providing a dynamical interpretation for this new quantity. Studying the unitary representations of the Poincaré group on the extended phase space allows us to define basis quantities for quantum states and develop the gauge invariant electromagnetic Hamiltonian in classical and quantum mechanics. We write plane wave solutions for free particles and construct stable singlet states. As in non-relativistic quantum theory the magnetic field couples to the rotation generators of SU(2) ⊂ SL(2, C), however the electric field couples to the non-compact and anti-Hermitian boost generators. We show that to first order in perturbation theory, a constant external magnetic field will produce a perturbative contribution to the total mass (the eigenvalue of the Lorentz scalar Hamiltonian) or a transition between states in the same representation of SL(2, C), and so will not disrupt a singlet state. However, an electric field normal to πµ can produce a perturbative transition between states in inequivalent representations of SL(2, C), and therefore potentially cause decoherence of the relativistic entanglement.