1Northwestern University Feinberg School of Medicine, Division of Nephrology and Hypertension, Chicago, Illinois 2Hospital Universitari Vall d’Hebron, Division of Nephrology Autonomous University of Barcelona, Barcelona, Spain 3Division of Nephrology, Department of Medicine, Duke University School of Medicine, Durham, North Carolina 4Renal Section, Durham Veterans Affairs Health Care System, Durham, North Carolina 5Division of Nephrology, Department of Medicine, University of Ottawa, Ottawa, Ontario, Canada 6Department of Pediatrics, Section of Nephrology, Wake Forest School of Medicine, Winston Salem, North Carolina 7School of Medicine, Departments of Medicine and Physiology, Johns Hopkins University, Baltimore, Maryland 8Division of Nephrology, and Center for Immunity, Inflammation and Regenerative Medicine, University of Virginia, Charlottesville, Virginia
Antonio Cammi, Lorenzo Loi, Andrea Missaglia
et al.
The global availability of high-intensity neutron sources is restricted by the prohibitive costs of spallation facilities and the decommissioning of aging research reactors, while compact accelerator-driven sources (CANS) are fundamentally limited by target power density and thermal-mechanical stress. Here, we introduce SYNERGY (SYnchrotron-driven NEutron source for Research, energy Generation and therapY), a paradigm-shifting architecture that overcomes these bottlenecks by decoupling charged-particle acceleration from neutron production. By utilizing a storage ring to drive external photoneutron targets via synchrotron radiation, this topological separation ensures targets interact exclusively with a continuous-wave (CW) photon beam, minimizing thermo-mechanical shocks and enabling beam powers exceeding 200 kW per beamline. Through a systematic parametric analysis cross-validated using OpenMC, MCNPX, and FLUKA, we demonstrate single-beamline neutron production rates from $2.8\times10^{14}$ n/s to $1.3\times10^{15}$ n/s. With an inherent multi-beamline capacity feeding up to 50 independent stations, the total facility intensity exceeds $6.0\times10^{16}$ n/s. By bridging the gap between laboratory and national-scale infrastructure, SYNERGY provides a high-intensity, multi-user platform for subcritical systems, medical isotope production, and boron neutron capture therapy.
Large language models (LLMs) often generate plausible yet incorrect answers, posing risks in safety-critical settings such as medicine. Human evaluation is expensive, and LLM-as-judge approaches risk introducing hidden errors. Recent white-box methods detect contextual hallucinations using model internals, focusing on the localization of the attention mass, but two questions remain open: do these approaches extend to predicting answer correctness, and do they generalize out-of-domains? We introduce Head Entropy, a method that predicts answer correctness from attention entropy patterns, specifically measuring the spread of the attention mass. Using sparse logistic regression on per-head 2-Renyi entropies, Head Entropy matches or exceeds baselines in-distribution and generalizes substantially better on out-of-domains, it outperforms the closest baseline on average by +8.5% AUROC. We further show that attention patterns over the question/context alone, before answer generation, already carry predictive signal using Head Entropy with on average +17.7% AUROC over the closest baseline. We evaluate across 5 instruction-tuned LLMs and 3 QA datasets spanning general knowledge, multi-hop reasoning, and medicine.
Emerging from NASA's spacecraft simulations in the 1960s, digital twin technology has advanced through industrial adoption to spark a healthcare transformation. A digital twin is a dynamic, data-driven virtual counterpart of a physical system, continuously updated through real-time data streams and capable of bidirectional interaction. In medicine, digital twin integrates imaging, biosensors, and computational models to generate patient-specific simulations that support diagnosis, treatment planning, and drug development. Representative applications include cardiac digital twin for predicting arrhythmia treatment outcomes, oncology digital twin for tracking tumor progression and optimizing radiotherapy, and pharmacological digital twin for accelerating drug discovery. Despite rapid progress, major challenges, including interoperability, data privacy, and model fidelity, continue to limit widespread clinical integration. Emerging solutions such as explainable AI, federated learning, and harmonized regulatory frameworks offer promising pathways forward. Looking ahead, advances in multi-organ digital twin, genomics integration, and ethical governance will be essential to ensure that digital twin shifts healthcare from reactive treatment to predictive, preventive, and truly personalized medicine.
Background:
Existing studies on anesthesiologists’ attitudes toward artificial intelligence (AI) leave a global understanding underexplored. This cross-sectional study aims to investigate Turkish anesthesiologists’ attitudes toward AI, examining its perceived benefits, limitations, and associated ethical concerns. Insights from this study aim to enhance understanding of AI’s role in anesthesiology within a cultural and ethical context.
Methods:
This nationwide study surveyed Turkish anesthesiologists. Descriptive statistics summarized categorical variables, Pearson’s Chi-square test compared variables between groups. Binary logistic regression analyzed associations between demographic factors and positive attitudes toward AI.
Results:
Among 293 valid responses, 69.6% of participants expressed positive attitudes toward AI. Gender (P = 0.01), employment setting (P < 0.001), and prior AI experience (P < 0.001) were significant predictors of positive attitudes. AI applications most frequently endorsed included preoperative assessments (93.1%), academic support (95.2%), and medical education (91.2%). Ethical concerns were prominent, with liability ambiguity (87.3%) and privacy issues (62.8%) being the most cited. Logistic regression revealed that participants aged 46–55 were significantly more likely to exhibit positive attitudes (OR = 3.744, P = 0.03), while those with over 15 years of experience were less likely to do so (OR = 0.105, P = 0.04).
Conclusions:
Turkish anesthesiologists exhibited predominantly positive attitudes toward AI, with prior experience playing a significant role in shaping perceptions. While AI was embraced for academic, educational, and noninvasive tasks, skepticism was present toward its application in invasive procedures. These findings highlight AI’s potential to enhance efficiency and patient safety while underscoring the need for comprehensive legal and ethical frameworks.
Randomized experiments are the gold standard for estimating treatment effects, and randomization serves as a reasoned basis for inference. In widely used stratified randomized experiments, randomization-based finite-population asymptotic theory enables valid inference for the average treatment effect, relying on normal approximation and a Neyman-type conservative variance estimator. However, when the sample size is small or the outcomes are skewed, the Neyman-type variance estimator may become overly conservative, and the normal approximation can fail. To address these issues, we propose a sharp variance estimator and two causal bootstrap methods to more accurately approximate the sampling distribution of the weighted difference-in-means estimator in stratified randomized experiments. The first causal bootstrap procedure is based on rank-preserving imputation and we prove its second-order refinement over normal approximation. The second causal bootstrap procedure is based on constant-treatment-effect imputation and is further applicable in paired experiments. In contrast to traditional bootstrap methods, where randomness originates from hypothetical super-population sampling, our analysis for the proposed causal bootstrap is randomization-based, relying solely on the randomness of treatment assignment in randomized experiments. Numerical studies and two real data applications demonstrate advantages of our proposed methods in finite samples. The \texttt{R} package \texttt{CausalBootstrap} implementing our method is publicly available.
Bastian Pfeifer, Christel Sirocchi, Marcus D. Bloice
et al.
In the realm of precision medicine, effective patient stratification and disease subtyping demand innovative methodologies tailored for multi-omics data. Clustering techniques applied to multi-omics data have become instrumental in identifying distinct subgroups of patients, enabling a finer-grained understanding of disease variability. This work establishes a powerful framework for advancing precision medicine through unsupervised random-forest-based clustering and federated computing. We introduce a novel multi-omics clustering approach utilizing unsupervised random-forests. The unsupervised nature of the random forest enables the determination of cluster-specific feature importance, unraveling key molecular contributors to distinct patient groups. Moreover, our methodology is designed for federated execution, a crucial aspect in the medical domain where privacy concerns are paramount. We have validated our approach on machine learning benchmark data sets as well as on cancer data from The Cancer Genome Atlas (TCGA). Our method is competitive with the state-of-the-art in terms of disease subtyping, but at the same time substantially improves the cluster interpretability. Experiments indicate that local clustering performance can be improved through federated computing.
The advent of foundation models (FMs) such as large language models (LLMs) has led to a cultural shift in data science, both in medicine and beyond. This shift involves moving away from specialized predictive models trained for specific, well-defined domain questions to generalist FMs pre-trained on vast amounts of unstructured data, which can then be adapted to various clinical tasks and questions. As a result, the standard data science workflow in medicine has been fundamentally altered; the foundation model lifecycle (FMLC) now includes distinct upstream and downstream processes, in which computational resources, model and data access, and decision-making power are distributed among multiple stakeholders. At their core, FMs are fundamentally statistical models, and this new workflow challenges the principles of Veridical Data Science (VDS), hindering the rigorous statistical analysis expected in transparent and scientifically reproducible data science practices. We critically examine the medical FMLC in light of the core principles of VDS: predictability, computability, and stability (PCS), and explain how it deviates from the standard data science workflow. Finally, we propose recommendations for a reimagined medical FMLC that expands and refines the PCS principles for VDS including considering the computational and accessibility constraints inherent to FMs.
Maxime Griot, Jean Vanderdonckt, Demet Yuksel
et al.
Large Language Models (LLMs) such as ChatGPT demonstrate significant potential in the medical domain and are often evaluated using multiple-choice questions (MCQs) modeled on exams like the USMLE. However, such benchmarks may overestimate true clinical understanding by rewarding pattern recognition and test-taking heuristics. To investigate this, we created a fictional medical benchmark centered on an imaginary organ, the Glianorex, allowing us to separate memorized knowledge from reasoning ability. We generated textbooks and MCQs in English and French using leading LLMs, then evaluated proprietary, open-source, and domain-specific models in a zero-shot setting. Despite the fictional content, models achieved an average score of 64%, while physicians scored only 27%. Fine-tuned medical models outperformed base models in English but not in French. Ablation and interpretability analyses revealed that models frequently relied on shallow cues, test-taking strategies, and hallucinated reasoning to identify the correct choice. These results suggest that standard MCQ-based evaluations may not effectively measure clinical reasoning and highlight the need for more robust, clinically meaningful assessment methods for LLMs.
G. Hamsaveni, A. M. Amrutha, Bhagyalaxmi Sidenur
et al.
Background:
Tuberculosis (TB) has significant health, social, psychological, and economic impacts on patients. This study aimed to assess and compare quality of life (QOL) among multidrug-resistant (MDR) TB patients and drug-sensitive TB patients.
MATERIALS AND METHODS:
A comparative study was conducted among 40 MDR-TB patients and 80 age- and gender-matched drug-sensitive TB patients in Chitradurga district, South India. Sociodemographic data were collected and QOL was assessed using the WHO BREF scale. Data were analyzed using SPSS v20.
Results:
The study population was predominantly male (65%) and aged 27–35 years. Education level and socioeconomic status differed significantly between MDR and sensitive TB groups (P < 0.05). MDR-TB patients had significantly lower self-rated QOL and health satisfaction compared to sensitive TB patients (P < 0.001). MDR-TB patients scored lower on all four QOL domains (physical, psychological, social relationships, and environmental), with the psychological domain most affected. The majority of both MDR and sensitive TB patients had “fair” QOL across domains.
Conclusion:
QOL was impaired in both MDR and sensitive TB patients, but more severely in MDR-TB. The psychological domain was most affected. Factors such as poor accessibility to health services, low socioeconomic status, and poor mental health may contribute to reduce QOL in TB patients.
Mohamed Sedky, Sherif Abd El Aziz, Shaaban Abd Elmoneum
et al.
Background: Coronavirus disease 2019 (COVID-19) is firstly reported in Wuhan, China. Then, it was quickly spread and becomes an epidemic. It is due to infection by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). It is highly transmissible with a great risk of mortality. Patients with diabetes mellitus (DM) are more prone to infectious agents like SARS-COV-2. Aim: The aim of the study is to evaluate the relationship between DM and COVID-19 infection regarding to its severity, mortality, rate of admission, complications, and prognosis.Patients and methods: A cross sectional study was performed between April 2021 and September 2021 and included 75 patients divided into two groups: Group A (COVID-19 patients with diabetes: n= 25), Group B (COVID-19 patients who developed diabetes: n= 25) and Group C (COVID-19 patients without diabetes: n= 25). Demographics, clinical, laboratory, radiologic, management, complications, and clinical outcomes data were collected and compared between the groups.Results: Patients with diabetes had a higher complication rate, like respiratory failure, acute cardiac injury. The respiratory failure did not significantly different between groups (it was 20%, 28% and 12% in groups A, B and C, respectively, P = .368). However, acute cardiac injury had been significantly increased in groups A than group B and in A and B than group C. (It was 44%, 20% and 8%, in groups A, B and C, respectively, P= 0.01). The mortality rate was also significantly higher among the A and B than C group (56%, 40% vs 8%, P=0.001).Conclusion: Diabetes is an independent risk factor for the prognosis of COVID-19. Diabetic patients should be intensely monitored during treatment, especially those who require insulin therapy.Background: Coronavirus disease 2019 [COVID-19] was first reported in Wuhan, China. It then rapidly spread and became a global epidemic due to infection by severe acute respiratory syndrome coronavirus 2 [SARS-CoV-2]. COVID-19 is highly transmissible with a high risk of mortality. Patients with diabetes mellitus [DM] are more susceptible to infectious agents like SARS-CoV-2.Aim of the work: The aim of the study was to evaluate the relationship between DM and COVID-19 infection regarding severity, mortality, admission rate, complications, and prognosis.Patients and Methods: A cross-sectional study was performed between April 2021 and September 2021. It included 75 patients divided into three groups: Group A [COVID-19 patients with diabetes, n=25], Group B [COVID-19 patients who developed diabetes, n=25] and Group C [COVID-19 patients without diabetes, n=25]. Demographic, clinical, laboratory, radiologic, management, complication, and clinical outcome data were collected and compared between the groups.Results: Patients with diabetes had a higher rate of complications like respiratory failure and acute cardiac injury. Respiratory failure was not significantly different between groups [20%, 28% and 12% in groups A, B and C respectively, P=0.368]. However, acute cardiac injury was significantly higher in groups A than B and in A and B than C [[44%, 20% and 8% respectively, P=0.01]. The mortality rate was also significantly higher among groups A and B than C [56%, 40% vs 8%, P=0.001].Conclusion: Diabetes is an independent risk factor for COVID-19 prognosis. Diabetic patients should be closely monitored during treatment, especially those requiring insulin therapy.
Abstract Objective To assess the dispositions, management, and clinical outcomes of TIAMS patients in ED to improve the quality of management in ED. Material and Method A descriptive retrospective study was conducted in ED patients aged >18 years diagnosed with TIAMS in the ED from 1 January 2018, to 31 January 2019. Data regarding terms of clinical presentation, examination, management, disposition, and adverse events were collected. Results Three hundred and sixty‐three TIAMS patients were enrolled in the study. Majority of the patients aged <45 years were admitted or referred (15.4%). The highest proportion of patients whose onset times from the last normal were less than 4.5 h were admitted to the EDOU (55.6%), while all patients whose onset times from the last normal were more than 48 h were discharged. Patients with abnormal cerebellar signs or atrial fibrillation were less likely to be discharged from the hospital. Patients with lower National Institutes of Health Stroke Scale (NIHSS) and ABCD2 scores tended to be discharged. Conclusion Among TIAMS patients, age, symptom onset, presence of atrial fibrillation, positive cerebellar signs, and severity scores influenced the disposition. There was no difference in adverse events among disposition groups.
Surgery, Medical emergencies. Critical care. Intensive care. First aid
Win Myat Phyo, Danuthida Saket, Marcio A. da Fonseca
et al.
Abstract Background Surface remineralization is recommended for the management of active non-cavitated interproximal carious lesions in primary teeth. According to the American Academy of Pediatric Dentistry, a recently recognized category of materials called bioactive restorative materials can be used for remineralization. This study aimed to evaluate the release of fluoride (F), calcium (Ca) and phosphate (P) ions from Predicta® Bioactive Bulk-fill composite compared with EQUIA Forte® and Filtek™ Z350 and to determine the remineralization effect of these 3 restorative materials on adjacent initial interproximal enamel carious lesions. Methods The release of F, Ca and P ions from 3 groups ((n = 10/group) (Group 1- Predicta®, Group 2- EQUIA Forte® and Group 3- Filtek™ Z350)) was determined at 1st, 4th, 7th and 14th days. After creating artificial carious lesions, human enamel samples were randomly assigned into 3 groups (n = 13/group) which were placed in contact with occluso-proximal restorative materials and exposed to a 14-day pH cycling period. Surface microhardness was determined using a Knoop microhardness assay at baseline, after artificial carious lesions formation and after pH cycling. The difference in the percentage of surface microhardness recovery (%SMHR) among groups was compared. Mineral deposition was analyzed with energy-dispersive x-ray spectroscopy (EDS) and the enamel surface morphology was evaluated with scanning electron microscopy (SEM). Kruskal-Wallis’s test with Dunn’s post hoc test and one-way ANOVA with Tukey’s post hoc test were used for data analysis. Results EQUIA Forte® released the highest cumulative amount of F and P ions, followed by Predicta® and Filtek™ Z350. Predicta® released higher amount of Ca ions than EQUIA Forte® and Filtek™ Z350. Predicta® demonstrated the highest %SMHR, followed by EQUIA Forte® and Filtek™ Z350. There was a significant difference in the %SMHR between Predicta® and Filtek™ Z350 (p < 0.05). However, EQUIA Forte® demonstrated the highest fluoride content, followed by Predicta® and Filtek™ Z350. The SEM images of EQUIA Forte® and Predicta® revealed the greater mineral deposition. Conclusion Predicta® demonstrated a marked increase in surface microhardness and fluoride content of adjacent initial interproximal enamel carious lesions in primary molars compared with Filtek™ Z350. Predicta® is an alternative restorative material to remineralize adjacent initial interproximal enamel carious lesions in primary molars, especially in high-risk caries patients.
ABSTRACT The interaction between coexisting plasmids can affect plasmid-carried resistance gene persistence and spread. However, whether the persistence of the blaCTX-M gene in clinical Enterobacteriaceae is related to the interaction of coresident nonresistance-conferring plasmids has not been reported. This study was initiated to elucidate how a nonresistance-conferring IncI1 plasmid affected the blaCTX-M-bearing IncFII plasmid colocated on the same cell. Herein, we constructed three isogenic derivatives of E. coli C600, designated as C600FII, C600I1, and C600FII+I1, which harbored the blaCTX-M-IncFII plasmid and/or the nonresistance-IncI1 one. We discovered that strain C600FII+I1 conferred higher fitness advantages than strain C600FII; also, the stability of the blaCTX-M-IncFII plasmid was noticeably improved in an antibiotic-free environment when it coexisted with the IncI1 plasmid. To further explore why the IncI1 plasmid enhanced the persistence of the blaCTX-M-IncFII plasmid, we assessed the blaCTX-M-IncFII plasmid's copy numbers, conjugation frequencies, and rep gene expressions in strains C600FII and C600FII+I1. The results demonstrated that the rep expressions of the blaCTX-M-IncFII plasmid in strain C600FII+I1 was greatly decreased, along with the plasmid’s copy numbers and mating efficiencies, compared to those in strain C600FII. Moreover, further study revealed that the intracellular ATP levels of strain C600FII+I1 were far lower than those of strain C600FII. Our findings confirmed that coexistence of the nonresistance-IncI1 plasmid can keep the blaCTX-M-IncFII plasmid more stable by increasing the fitness advantages of the host bacteria, which will pose a threat to preventing the long-term presence of the plasmid-carried blaCTX-M gene in clinical Enterobacteriaceae.IMPORTANCESo far, plasmid-carried blaCTX-M is still the most common extended-spectrum beta-lactamase (ESBL) genotype in clinical settings worldwide. Except for the widespread use of third-generation cephalosporins, the interaction between coexisting plasmids can also affect the long-term stable existence of the blaCTX-M gene; however, the study on that is still sparse. In the present study, we assess the interaction of coinhabitant plasmids blaCTX-M-IncFII and nonresistance-IncI1. Our results confirmed that the increased fitness advantages of strain C600FII+I1 were attributable to the cohabitant nonresistance-IncI1 plasmid, which largely reduced the intracellular ATP levels of host bacteria, thus decreasing the rep gene expression of the blaCTX-M-IncFII plasmid, its copy numbers, and mating efficiencies, while the higher fitness advantages of strain C600FII+I1 enhanced the persistence of the blaCTX-M-IncFII plasmid. The results indicate that the nonresistance-IncI1 plasmid contributes to the long-term existence of the blaCTX-M-IncFII plasmid, implying a potentially new strategy for controlling the spread of resistance plasmids in clinical settings by targeting nonresistance plasmids.
Kayoko Ishii, Hiroko Sumita, Hitomi Nagamine
et al.
Abstract Background In occupational health, the maintenance and promotion of workers’ health, especially lifestyle motivation-based interventions, have gained considerable attention and are actively implemented. Motivational theories include self-determination theory, and some studies focus on healthy lifestyles. However, the effectiveness of health promotion interventions varies depending on the health awareness and motivation of the participants. Therefore, this study aimed to clarify the processes by which workers are motivated to improve their health and to identify the need for and type of support according to their motivation. Methods Using a mixed-research design, an initial questionnaire survey of 94 employees (mean age = 40.97 ± 9.65) at a multicenter company in Japan, followed by semi-structured interviews with 16 employees (mean age = 40.13 ± 9.45) from the high- and low-motivation groups, were conducted. Multiple regression analysis followed by modified grounded theory-based analysis of the results of the first stage was used and the quantitative and qualitative results were integrated. Results In the first stage, autonomous motivation scores were predicted by the behavioral change stage and relatedness satisfaction/frustration. The second stage revealed that “the process of reflecting and managing one’s own health while receiving support and feedback for maintaining and improving health” was the motivational process of workers. Result integration revealed that motivation increased through repeatedly escaping and adjusting to real problems and situational coping until the behavioral change. Despite interruptions during behavioral change, receiving feedback from others could increase motivation and continued behavioral change. Conclusion Regardless of their level of motivation for health behaviors, workers indicated that support from others was essential. The nature of this support was found to range from providing information to offering feedback. Interventions individualized by the identified process could enable customized motivation-driven health guidance.
Andrea Schillaci, Kazuto Hasegawa, Carlotta Pipolo
et al.
In several problems involving fluid flows, Computational Fluid Dynamics (CFD) provides detailed quantitative information, and often allows the designer to successfully optimize the system, by minimizing a cost function. Sometimes, however, one cannot improve the system with CFD alone, because a suitable cost function is not readily available: one notable example is diagnosis in medicine. The field of interest considered here is rhinology: a correct air flow is key for the functioning of the human nose, yet the notion of a functionally normal nose is not available, and a cost function cannot be written. An alternative and attractive pathway to diagnosis and surgery planning is offered by data-driven methods. In this work, we consider the machine-learning study of nasal pathologies caused by anatomic malformations, with the aim of understanding whether fluid dynamic features, available after a CFD analysis, are more effective than purely geometric features in the training of a neural network for regression. Our experiments are carried out on an extremely simplified anatomic model and a correspondingly simple CFD approach; nevertheless, they demonstrate that flow-based features perform better than geometry-based ones, and allow the training of a neural network with fewer inputs, a crucial advantage in fields like medicine.
Andreas Triantafyllopoulos, Alexander Kathan, Alice Baird
et al.
Recent years have seen a rapid increase in digital medicine research in an attempt to transform traditional healthcare systems to their modern, intelligent, and versatile equivalents that are adequately equipped to tackle contemporary challenges. This has led to a wave of applications that utilise AI technologies; first and foremost in the fields of medical imaging, but also in the use of wearables and other intelligent sensors. In comparison, computer audition can be seen to be lagging behind, at least in terms of commercial interest. Yet, audition has long been a staple assistant for medical practitioners, with the stethoscope being the quintessential sign of doctors around the world. Transforming this traditional technology with the use of AI entails a set of unique challenges. We categorise the advances needed in four key pillars: Hear, corresponding to the cornerstone technologies needed to analyse auditory signals in real-life conditions; Earlier, for the advances needed in computational and data efficiency; Attentively, for accounting to individual differences and handling the longitudinal nature of medical data; and, finally, Responsibly, for ensuring compliance to the ethical standards accorded to the field of medicine.