Hasil untuk "Internal medicine"

R. Mittal, R. Holloway, R. Penagini et al.

R. Haynes, N. Wilczynski, K. McKibbon et al.

Tatsuaki Tsuruyama

We consider a response system that updates its internal state in accordance with information input arriving from outside. In this paper, we define as internal time the ``number of kinds'' of codes that have been observed at least once up to a given time, and analyze how the way internal time advances is determined by the statistics of information input (arrival rate and code distribution). When arrivals follow a Poisson process, the average advancing speed of internal time decreases monotonically with time, and if the number of kinds of codes is finite, it eventually approaches an upper limit and saturates. As a result, on long time scales, internal time becomes relatively shorter than physical time. For a uniform code distribution, we provide a closed form for the correspondence between internal time and physical time, and show that the physical time required to ``advance internal time by one step'' increases in later stages. As an ancillary quantity, we quantify by conditional entropy the remaining uncertainty of ``which codes have been observed'' when only internal time is known, and we give unimodality and the maximization time in the uniform case, and upper bounds, equality conditions, and expressions of the difference from the upper bound in the non-uniform case. Finally, we also present a generalization that assigns weights (description lengths) to each code so that internal time is ticked according to the amount of information in the input.

J. Curtis, A. Back, D. Ford et al.

G. Ruiz‐Irastorza, M. Cuadrado, I. Ruiz-Arruza et al.

Xin Zhou, Yiwen Guo, Ruotian Ma et al.

Aligning Large Language Models (LLMs) with human preferences is crucial for their deployment in real-world applications. Recent advancements in Self-Rewarding Language Models suggest that an LLM can use its internal reward models (such as LLM-as-a-Judge) \cite{yuanself} to generate preference data, improving alignment performance without costly human annotation. However, we find that different internal reward models within the same LLM often generate inconsistent preferences. This inconsistency raises concerns about the reliability of self-generated preference data, hinders overall alignment performance, and highlights the need for further research to ensure reliable and coherent alignment with human preferences. To address this limitation, we propose Self-Consistent Internal Rewards (SCIR), a novel framework designed to enhance consistency among internal reward models during training. In each training step, we collect preference predictions from multiple pre-defined internal reward models and enforce consistency and confidence through an inconsistency penalty mechanism, thereby improving the reliability of these internal reward models. We selectively use data with consistent predictions for preference optimization, ensuring the quality of the preference data. By employing self-consistent internal rewards, our method significantly improves the alignment performance and reward modeling capability of LLMs, outperforming baseline methods by a notable margin.

Syed Ameen Ahmad, Olivia Liu, Amy Feng et al.

Abstract Background There is an emerging understanding of the increased risk of stroke in patients with immune thrombocytopenic purpura (ITP) and immune thrombotic thrombocytopenic purpura (iTTP). We aimed to determine the prevalence and characteristics of acute ischemic stroke (AIS) and intracranial hemorrhage (ICH) in patients with ITP and iTTP in a systematic review and meta-analysis. Methods We used PubMed, Embase, Cochrane, Web of Science, and Scopus using text related to ITP, iTTP, stroke, AIS, and ICH from inception to 11/3/2023. Our primary outcome was to determine prevalence of AIS and/or ICH in a cohort of ITP or iTTP patients (age > 18). Our secondary outcomes were to determine stroke type associated with thrombopoietin receptor agonists (TPO-RAs) in ITP patients, as well as risk factors associated with stroke in ITP and iTTP patients. Results We included 42 studies with 118,019 patients (mean age = 50 years, 45% female). Of those, 27 studies (n = 116,334) investigated stroke in ITP patients, and 15 studies (n = 1,685) investigated stroke in iTTP patients. In all ITP patients, the prevalence of AIS and ICH was 2.1% [95% Confidence Interval (CI) 0.8-4.0%] and 1.5% (95% CI 0.9%-2.1%), respectively. ITP patients who experienced stroke as an adverse event (AE) from TPO-RAs had an AIS prevalence of 1.8% (95% CI 0.6%-3.4%) and an ICH prevalence of 2.0% (95% CI 0.2%-5.3%). Prevalence of stroke did not significantly differ between all ITP patients and those treated with TPO-RAs. iTTP patients had a prevalence of AIS and ICH of 13.9% (95% CI 10.2%-18.1%) and 3.9% (95% CI 0.2%-10.4%), respectively. Subgroup analysis revealed the prevalence of AIS and ICH was greater in iTTP patients vs. all ITP patients (p < 0.01 and p = 0.02, respectively). Meta-regression analysis revealed none of the collected variables (age, sex, history of diabetes or hypertension) were risk factors for stroke in all ITP patients, although there were high levels of data missingness. Conclusions Prevalence of different stroke types was lower in all ITP patients vs. iTTP patients. Additionally, ITP patients experienced a similar prevalence of stroke regardless of if they were specifically denoted to have been treated with TPO-RAs or not, supporting the continued use of TPO-RAs in management. Risk factors for stroke remain unclear, and future studies should continue to investigate this relationship.

Rudzani Muloiwa

Infection remains one of the greatest killers of children in the world. A large proportion of these deaths are due to infectious agents for which effective vaccines are already available.Vaccination has been one the most successful and cost-effective public health interventions of the 20th century. Data indicates that the greatest return on investment from vaccination disproportionately accrues to the poorest communities. Unfortunately, access to vaccines is not equitable, with the most vulnerable least likely to receive them.Using global data on the burden of disease, vaccination coverage, historical patterns of vaccine introduction and uptake, as well as access to currently available antigens, the talk will highlight the inequities in immunization practices and explore their causes. In addition, the talk will challenge the ethical framework that perpetuates these iniquities and suggests potential opportunities for addressing them.

J. Apelqvist, K. Bakker, W. H. Houtum et al.

K. Schulz, Douglas G. Altman, David Moher

The CONSORT statement is used worldwide to improve the reporting of randomised controlled trials. Kenneth Schulz and colleagues describe the latest version, CONSORT 2010, which updates the reporting guideline based on new methodological evidence and accumulating experience.To encourage dissemination of the CONSORT 2010 Statement, this article is freely accessible on bmj.com and will also be published in the Lancet, Obstetrics and Gynecology, PLoS Medicine, Annals of Internal Medicine, Open Medicine, Journal of Clinical Epidemiology, BMC Medicine, and Trials.

K. Templeton, C. Bernstein, J. Sukhera et al.

Kim Templeton, MD, University of Kansas Medical Center; Carol A. Bernstein, MD, Albert Einstein College of Medicine and Montefi ore Health; Javeed Sukhera, MD, PhD, FRCPC, Western University Canada and London Health Sciences Centre; Lois Margaret Nora, MD, JD, MBA, Northeast Ohio Medical University; Connie Newman, MD, New York University School of Medicine; Helen Burstin, MD, MPH, Council of Medical Specialty Societies; Constance Guille, MD, Medical University of South Carolina; Lorna Lynn, MD, American Board of Internal Medicine; Margaret L. Schwarze, MD, MPP, FACS, University of Wisconsin School of Medicine and Public Health; Srijan Sen, MD, PhD, University of Michigan; and Neil Busis, MD, University of Pittsburgh School of Medicine and UPMC Shadyside

U. Pape, A. Perren, B. Niederle et al.

Ju-Yeon Cho, Young-Sun Lee, S. Kim et al.

1 Department of Internal Medicine, College of Medicine, Chosun University, Gwangju, Korea; Department of Internal Medicine, Guro Hospital, Korea University College of Medicine, Seoul, Korea; 3 Department of Gastroenterology, Ajou University School of Medicine, Suwon, Korea; 4 St. Vincent’s Hospital, The Catholic University of Korea, Seoul, Korea; 5 Department of Internal Medicine and Liver Research Institute, Seoul National University College of Medicine.

Guangzhi Xiong, Qiao Jin, Xiao Wang et al.

The emergent abilities of large language models (LLMs) have demonstrated great potential in solving medical questions. They can possess considerable medical knowledge, but may still hallucinate and are inflexible in the knowledge updates. While Retrieval-Augmented Generation (RAG) has been proposed to enhance the medical question-answering capabilities of LLMs with external knowledge bases, it may still fail in complex cases where multiple rounds of information-seeking are required. To address such an issue, we propose iterative RAG for medicine (i-MedRAG), where LLMs can iteratively ask follow-up queries based on previous information-seeking attempts. In each iteration of i-MedRAG, the follow-up queries will be answered by a conventional RAG system and they will be further used to guide the query generation in the next iteration. Our experiments show the improved performance of various LLMs brought by i-MedRAG compared with conventional RAG on complex questions from clinical vignettes in the United States Medical Licensing Examination (USMLE), as well as various knowledge tests in the Massive Multitask Language Understanding (MMLU) dataset. Notably, our zero-shot i-MedRAG outperforms all existing prompt engineering and fine-tuning methods on GPT-3.5, achieving an accuracy of 69.68% on the MedQA dataset. In addition, we characterize the scaling properties of i-MedRAG with different iterations of follow-up queries and different numbers of queries per iteration. Our case studies show that i-MedRAG can flexibly ask follow-up queries to form reasoning chains, providing an in-depth analysis of medical questions. To the best of our knowledge, this is the first-of-its-kind study on incorporating follow-up queries into medical RAG. The implementation of i-MedRAG is available at https://github.com/Teddy-XiongGZ/MedRAG.

Bohuslav Matouš

This paper connects two methods for finding the functional of entropy in F(R)-Gravity: Padmanabhan's and Hammad's. The resulting approach is simple to follow and yields entropy functional, which can be separated into two parts. The part unknown in General Relativity is often called in the literature as an internal entropy and this paper points on incompatibility between the internal entropy found from the entropy functional and the one found using conventional approach.

Pinak Mandal, Georg A. Gottwald, Nicholas Cranch

The computationally cheap machine learning architecture of random feature maps can be viewed as a single-layer feedforward network in which the weights of the hidden layer are random but fixed and only the outer weights are learned via linear regression. The internal weights are typically chosen from a prescribed distribution. The choice of the internal weights significantly impacts the accuracy of random feature maps. We address here the task of how to best select the internal weights. In particular, we consider the forecasting problem whereby random feature maps are used to learn a one-step propagator map for a dynamical system. We provide a computationally cheap hit-and-run algorithm to select good internal weights which lead to good forecasting skill. We show that the number of good features is the main factor controlling the forecasting skill of random feature maps and acts as an effective feature dimension. Lastly, we compare random feature maps with single-layer feedforward neural networks in which the internal weights are now learned using gradient descent. We find that random feature maps have superior forecasting capabilities whilst having several orders of magnitude lower computational cost.

Hubert Daisley, Oneka Acco, Martina Daisley et al.

The vasa vasorum of the large pulmonary vessels is involved in the pathology of COVID-19. This specialized microvasculature plays a major role in the biology and pathology of the pulmonary vessel walls. We have evidence that thrombosis of the vasa vasorum of the large and medium-sized pulmonary vessels during severe COVID-19 causes ischemia and subsequent death of the pulmonary vasculature endothelium. Subsequent release of thrombi from the vasa interna into the pulmonary circulation and pulmonary embolism generated at the ischemic pulmonary vascular endothelium site, are the central pathophysiological mechanisms in COVID-19 responsible for pulmonary thromboembolism. The thrombosis of the vasa vasorum of the large and medium-sized pulmonary vessels is an internal event leading to pulmonary thromboembolism in COVID-19.

Han Qiao, Yan Feng, Xiaolei Han et al.

BackgroundThis study focuses on determining the prognostic and predictive value of the comprehensive prognostic nutrition index (FIDA) in individuals undergoing treatment for Non-Small-Cell Lung Carcinoma (NSCLC).MethodsThis retrospective analysis encompassed 474 of NSCLC patients treated from January 2010 through December 2019. Employing the Lasso-COX regression approach, eight blood parameters were identified as significant prognostic indicators. These parameters contributed to the formulation of the comprehensive prognostic nutrition index FIDA. Utilizing X-tile software, the patient cohort was categorized into either a high or low FIDA group based on an established optimal threshold. The cohort was then randomly segmented into a training set and a validation set using SPSS software. Subsequent steps involved conducting univariate and multivariate regression analyze to develop a prognostic nomogram. The effectiveness of this nomogram was evaluated by calculating the AUC.ResultsAnalysis of survival curves for both the training and validation sets revealed a poorer prognosis in the high FIDA group compared to the low FIDA group. This trend persisted across various subgroups, including gender, age, and smoking history, with a statistical significance (p&lt;0.05). Time-dependent ROC and diagnostic ROC analyses affirmed that FIDA serves as an effective diagnostic and prognostic marker in NSCLC. Moreover, Cox regression multivariate analysis established FIDA as an independent prognostic factor for NSCLC. The prognostic nomogram, integrating FIDA and clinical data, demonstrated substantial prognostic utility and outperformed the traditional TNM staging systemin predicting overall survival (OS).ConclusionFIDA emerges as a dependable predictor of outcomes for patients with NSCLC. It offers a practical, cost-effective tool for prognostication in regular clinical applications.

Paraskevi Farmaki, C. Damaskos, N. Garmpis et al.

First Department of Pediatrics, Aghia Sophia Children's Hospital, Athens, Greece; Second Department of Propedeutic Surgery, Laiko General Hospital, Medical School, National and Kapodistrian University of Athens, Athens, Greece; Internal Medicine Department, Laiko General Hospital, Medical School, National and Kapodistrian University of Athens, Athens, Greece; Department of Internal Medicine, General Hospital of Athens "Elpis", Athens, Greece; Health Center Peristeriou, Athens, Greece

Jingzi Huang, Henry C. Burridge, Maarten van Reeuwijk

We study the mixing processes inside a forced fountain using data from direct numerical simulation. The outer boundary of the fountain with the ambient is a turbulent/non-turbulent interface. Inside the fountain, two internal boundaries, both turbulent/turbulent interfaces, are identified: 1) the classical boundary between upflow and downflow which is composed of the loci of points of zero mean vertical velocity; and 2) the streamline that separates the mean flow emitted by the source from the entrained fluid from the ambient (the separatrix). We show that entrainment due to turbulent fluxes across the internal boundary is at least as important as that by the mean flow. However, entrainment by the turbulence behaves substantively differently from that by the mean flow and cannot be modelled using the same assumptions. This presents a challenge for existing models of turbulent fountains and other environmental flows that evolve inside turbulent environments.