The introduction of network science approaches into public transport research has seen great advances in the past 15 years. However, it has become apparent that monolayer networks are often not sufficient to model and analyse real-world systems in sufficient detail. In the last decade, the theory of multilayer networks has proven to be an invaluable tool in various disciplines, including transport. Multilayer networks consist of layers of networks that are coupled among themselves. This enables modelling of complex systems with heterogeneous elements and relations between them. Although there is a body of work in public transport research that uses multilayer networks, the related literature is scattered, lacking unified terminology and agreed-upon approaches. We posit that there is vast uncovered potential in using multilayer network approaches to public transport modelling, planning, and operations. We first present the basic formalisms of multilayer networks with a focus on how they (may) relate to public transport networks. We then provide a systematic review of the literature on multilayer networks in public transport research. We identify and taxonomise ways in which public transport systems are modelled as multilayer networks. Based on the survey and drawing from the state and history of network science in public transport research as well as multilayer approaches across other application domains, we propose a research agenda for multilayer public transport networks for the upcoming decade(s).
The role of technical regulation during import of production products is difficult to overestimate since it is just on quality of tests carried out regarding goods for conformity with requirements of normative documents depends safety of consumer health. Nowadays, technical regulation system is undergoing large-scale changes related to digitization and application of risk-oriented approach to selection of objects for control that undoubtedly will improve situation with falsification of test protocols. However, there is still impressive percentage of production that do not meet the regulated requirements. So, in production destined for children, after verification activities, discrepancies related to chemical, physical and toxicological parameters were found. The exceeded standards content of heavy metals, phthalates and formaldehyde can result into such irreversible negative effects on health of socially vulnerable consumers as allergic reactions, intoxications, derangement of hormonal status, disorders of kidney and digestive system. It is very important to eliminate problematic aspects of technical regulation system with purpose of non-admission to inner market production that can be unsafe for health of citizens. The article presents recommendations for improving technical regulation system both from point of view of supervisory functions and differentiation and hardening of sanctions with respect to testing laboratories.
Hasti Narimanzadeh, Arash Badie-Modiri, Iuliia Smirnova
et al.
Affective polarization and political sorting drive public antagonism around issues at the science-policy nexus. Looking at the COVID-19 period, we study cross-domain spillover of incivility and contentiousness in public engagements with climate change and public health on Twitter and Reddit. We find strong evidence of the signatures of affective polarization surrounding COVID-19 spilling into the climate change domain. Across different social media systems, COVID-19 content is associated with incivility and contentiousness in climate discussions. These patterns of increased antagonism were responsive to pandemic events that made the link between science and public policy more salient. The observed spillover activated along pre-pandemic political cleavages, specifically anti-internationalist populist beliefs, that linked climate policy opposition to vaccine hesitancy. Our findings show how affective polarization in public engagement with science becomes entrenched across science policy domains.
Nesrine Fourati, Alisa Barkar, Marion Dragée
et al.
Background: Public speaking is a vital professional skill, yet it remains a source of significant anxiety for many individuals. Traditional training relies heavily on expert coaching, but recent advances in AI has led to novel types of commercial automated public speaking feedback tools. However, most research has focused on prototypes rather than commercial applications, and little is known about how public speaking experts perceive these tools. Objectives: This study aims to evaluate expert opinions on the efficacy and design of commercial AI-based public speaking training tools and to propose guidelines for their improvement. Methods: The research involved 16 semi-structured interviews and 2 focus groups with public speaking experts. Participants discussed their views on current commercial tools, their potential integration into traditional coaching, and suggestions for enhancing these systems. Results and Conclusions: Experts acknowledged the value of AI tools in handling repetitive, technical aspects of training, allowing coaches to focus on higher-level skills. However they found key issues in current tools, emphasising the need for personalised, understandable, carefully selected feedback and clear instructional design. Overall, they supported a hybrid model combining traditional coaching with AI-supported exercises.
Arti Virkud, Jessie K. Edwards, Michele Jonsson Funk
et al.
Identifying optimal medical treatments to improve survival has long been a critical goal of pharmacoepidemiology. Traditionally, we use an average treatment effect measure to compare outcomes between treatment plans. However, new methods leveraging advantages of machine learning combined with the foundational tenets of causal inference are offering an alternative to the average treatment effect. Here, we use three unique, precision medicine algorithms (random forests, residual weighted learning, efficient augmentation relaxed learning) to identify optimal treatment rules where patients receive the optimal treatment as indicated by their clinical history. First, we present a simple hypothetical example and a real-world application among heart failure patients using Medicare claims data. We next demonstrate how the optimal treatment rule improves the absolute risk in a hypothetical, three-modifier setting. Finally, we identify an optimal treatment rule that optimizes the time to outcome in a real-world heart failure setting. In both examples, we compare the average time to death under the optimized, tailored treatment rule with the average time to death under a universal treatment rule to show the benefit of precision medicine methods. The improvement under the optimal treatment rule in the real-world setting is greatest (additional ~9 days under the tailored rule) for survival time free of heart failure readmission.
Objective To develop an LLM based realtime compound diagnostic medical AI interface and performed a clinical trial comparing this interface and physicians for common internal medicine cases based on the United States Medical License Exam (USMLE) Step 2 Clinical Skill (CS) style exams. Methods A nonrandomized clinical trial was conducted on August 20, 2024. We recruited one general physician, two internal medicine residents (2nd and 3rd year), and five simulated patients. The clinical vignettes were adapted from the USMLE Step 2 CS style exams. We developed 10 representative internal medicine cases based on actual patients and included information available on initial diagnostic evaluation. Primary outcome was the accuracy of the first differential diagnosis. Repeatability was evaluated based on the proportion of agreement. Results The accuracy of the physicians' first differential diagnosis ranged from 50% to 70%, whereas the realtime compound diagnostic medical AI interface achieved an accuracy of 80%. The proportion of agreement for the first differential diagnosis was 0.7. The accuracy of the first and second differential diagnoses ranged from 70% to 90% for physicians, whereas the AI interface achieved an accuracy rate of 100%. The average time for the AI interface (557 sec) was 44.6% shorter than that of the physicians (1006 sec). The AI interface ($0.08) also reduced costs by 98.1% compared to the physicians' average ($4.2). Patient satisfaction scores ranged from 4.2 to 4.3 for care by physicians and were 3.9 for the AI interface Conclusion An LLM based realtime compound diagnostic medical AI interface demonstrated diagnostic accuracy and patient satisfaction comparable to those of a physician, while requiring less time and lower costs. These findings suggest that AI interfaces may have the potential to assist primary care consultations for common internal medicine cases.
Background:
Stress is associated with the development and worsening of depression disorder. Therefore, various stress reduction interventions have been surged to cope with stress and enhance treatment outcomes for patients with depressive disorder. The present study aims to evaluate the effect of nurse-led de-stressful skill interventions (NDSI) among patients with depression.
Materials and Methods:
Patients with depression disorders aged between 18 and 65 years will be randomly allocated into two groups. The treatment-as-usual (TAU) group patients will receive standard care. In contrast, patients in the NDSI group will receive seven sessions of the NDSI interventions over 8 weeks. Primary outcomes are stress, resilience, self-esteem, coping strategies, and quality of life. Data will be collected at three different intervals: baseline, post-intervention at 8 weeks, and follow-up at 12 weeks. The following standardized instruments will be applied: Perceived Stress Scale, Brief Resilience Scale, Rosenberg Self-esteem Scale, Brief COPE Scale, Quality of Life Enjoyment and Satisfaction Questionnaire Short Form, and Hamilton Depression Rating Scale.
Conclusion:
The finding of this study is expected to determine the effect of NDSI on patients with depression disorder, which would optimize treatment in patients with depression disorder.
Special aspects of education, Public aspects of medicine
Jakub Fusiak, Andreas Wolkenstein, Verena S. Hoffmann
BackgroundPatient preferences are a critical component of shared decision-making (SDM), particularly when choosing between treatment options with differing risks and outcomes. Many methods exist to elicit these preferences, but their complexity, usability, and acceptance vary.ObjectiveWe aim to gain insight into the acceptance, effort and preferences of participants regarding five different methods of preference assessment. Additionally, we investigate the influence of health status, experiences within the health system and of demographic factors on the results.MethodsWe conducted a cross-sectional online survey including five preference elicitation Methods: best-worst scaling, direct weighting, PAPRIKA (Potentially All Pairwise Rankings of all Possible Alternatives), time trade-off, and standard gamble. The questionnaire was distributed via academic and patient advocacy mailing lists, reaching both healthy individuals and those with acute or chronic illnesses. Participants rated each method using six standardized statements on a 5-point Likert scale. Additional items assessed general acceptance of algorithm-assisted preference assessments and the clarity of the questionnaire.ResultsOf 258 initiated questionnaires, 123 (48%) were completed and included in the analysis. Participants were diverse in age, gender, and health status, but predominantly highly educated and digitally literate. Across all measures, the PAPRIKA method received the highest ratings for clarity, usability, and perceived ability to express preferences. Simpler methods (best-worst scaling, direct weighting) were rated as less useful for capturing nuanced preferences, while abstract utility-based methods (standard gamble, time trade-off) were seen as cognitively demanding. Subgroup analyses showed minimal variation across demographic groups. Most participants (82%) could imagine using at least one of the presented methods in real clinical settings, but also emphasized the importance of physician involvement in interpreting results.ConclusionThe interactive PAPRIKA method best balanced cognitive demand and expressiveness and was preferred by most participants. Structured methods for preference elicitation may enhance SDM when integrated into clinical workflows and supported by healthcare professionals. Further research is needed to evaluate their use in real-world decisions and among more diverse patient populations.
Nick Bansback, Mina Tadrous, Marie Paul Nisingizwe
et al.
Background The introduction of direct-acting antivirals (DAAs) has allowed countries to reduce the health and economic burden of hepatitis C virus (HCV). However, access to DAAs remains expensive and limited in many countries globally due to wide disparities in HCV drug pricing. We assessed how global use of HCV drugs has changed over time and the effect that COVID-19 might have had on DAA utilisation.Methods We assessed longitudinal changes in DAA sales by country income group, geographical region and drug type. We also conducted an interrupted time series analysis to assess COVID-19-related changes in the trend of DAA units sold globally. Our analysis used DAA sales data from the IQVIA multinational integrated data analysis database of 52 countries and two regions and HCV prevalence data from Polaris from 2014 to 2020. Our primary outcome was the monthly rate of DAAs sold per 100 000 people living with HCV per country, country income group and geographic region. We then compared the pre-post change in DAA units by drug type and country income group. We fitted autoregressive moving average models with a ramp function to assess the impact of COVID-19 on monthly DAA units sold.Results Across all countries, from August 2014 to August 2020, a monthly average of 44 219 DAA units per 100 000 HCV cases was sold. High-income countries purchased more units than other groups. In terms of geographic location, North America (124 144 per 100 000 HCV cases) and Europe (81 001 per 100 000 HCV cases) had the highest DAA sales over time; the newer generation of combination DAAs was mainly used in high-income countries. In contrast, first-generation and second-generation DAAs were the predominant types of DAAs sold in lower middle-income countries (LMICs). The pre-post analysis showed a 23% (p<0.001) average decrease in global sales of DAAs during the first phase of COVID-19. The decrease in LMICs (69%, p<0.001) was approximately double that of high-income countries (33%, p<0.001), while upper middle-income countries (UMICs) had a 34% (p<0.001) increase in DAA sales. The pandemic was associated with an immediate and sustained decrease of 9263 units per month (95% CI −14 668 to −3857.46) in high-income countries, a 73.14 (−850.96 to 997.24) unit increase in UMICs and a 742.58 (95% CI −5505.91 to 4020.75) unit decrease in LMICs.Conclusion Our study showed uneven access to DAAs globally, with higher prevalence-adjusted utilisation in high-income countries compared with lower-income countries. Our study also found that the COVID-19 pandemic has significantly decreased DAA sales in many countries. To counter these trends, additional strategies, such as price reductions, increased competition among manufacturers and licensing agreements, may help to improve access and utilisation of DAAs globally.
This article substantiates the relevance of ethical understanding and regulatory restrictions on the process of implementing a system of cognitive expert systems in healthcare. The key provisions of the axiological approach to digital healthcare strategies are defined, which should take into account national and regional conditions with mandatory dynamic interaction of all stakeholders, guaranteeing, among other things, international pluralism and basic community. Based on the analysis of a number of systems, it was revealed that the main thing in the process of developing and using expert systems is the correctness, reliability of medical information and professional assessment of the results of the systems activities. Which not only does not contradict, but also accepts the solution to the problem with the mandatory systematic inclusion of basic ethical standards in the process of developing technical solutions based on artificial intelligence and active parallel legislative initiative.
Davide Belluomo, Tiziana Calamoneri, Giacomo Paesani
et al.
We present a new unified graph-based representation of medical data, combining genetic information and medical records of patients with medical knowledge via a unique knowledge graph. This approach allows us to infer meaningful information and explanations that would be unavailable by looking at each data set separately. The systematic use of different databases, managed throughout the built knowledge graph, gives new insights toward a better understanding of oncology medicine. Indeed, we reduce some useful medical tasks to well-known problems in theoretical computer science for which efficient algorithms exist.
While large language models (LLMs) have achieved state-of-the-art performance on a wide range of medical question answering (QA) tasks, they still face challenges with hallucinations and outdated knowledge. Retrieval-augmented generation (RAG) is a promising solution and has been widely adopted. However, a RAG system can involve multiple flexible components, and there is a lack of best practices regarding the optimal RAG setting for various medical purposes. To systematically evaluate such systems, we propose the Medical Information Retrieval-Augmented Generation Evaluation (MIRAGE), a first-of-its-kind benchmark including 7,663 questions from five medical QA datasets. Using MIRAGE, we conducted large-scale experiments with over 1.8 trillion prompt tokens on 41 combinations of different corpora, retrievers, and backbone LLMs through the MedRAG toolkit introduced in this work. Overall, MedRAG improves the accuracy of six different LLMs by up to 18% over chain-of-thought prompting, elevating the performance of GPT-3.5 and Mixtral to GPT-4-level. Our results show that the combination of various medical corpora and retrievers achieves the best performance. In addition, we discovered a log-linear scaling property and the "lost-in-the-middle" effects in medical RAG. We believe our comprehensive evaluations can serve as practical guidelines for implementing RAG systems for medicine.
This is a series of three lectures I gave at the Korea Institute of Advanced Study in June 2019 at a workshop about "Algebraic and Symplectic Aspects of Degenerations of Complex Surfaces". I focus on the symplectic aspects, in particular on the case of cyclic quotient surface singularities. These notes have been available on a public Git repository since 2019, and I noticed that people occasionally cited them in the years since. For that reason, I decided to post them on arXiv for a more permanent record; I have made some small corrections and annotations but otherwise they are unchanged. These notes are a purely expository account of stuff I was thinking about 2016-2019, and are largely self-aggrandising.
Monica M. Diaz, Lesley A. Guareña, Bettsie Garcia
et al.
Summary: Goals of care (Goals-of-care) discussions and palliative care (PC) are crucial to providing comprehensive healthcare, particularly for acute neurological conditions requiring admission to a neurological intensive care unit. We identified gaps in the literature and describe insight for future research on end-of-life discussions and PC for U.S. Latinos with acute neurological conditions. We searched 10 databases including peer-reviewed abstracts and manuscripts of hospitalized U.S. Latinos with acute neurological and non-neurological conditions. We included 44 of 3231 publications and identified various themes: PC utilization, pre-established advanced directives in Goals-of-care discussions, Goals-of-care discussion outcomes, tracheostomy or percutaneous gastrostomy tube placement rates among hospitalized Latinos. Our review highlights that Latinos appear to have lower palliative care utilization compared with non-Latino Whites and may be less likely to have pre-established advanced directives, more likely to have gastrostomy or tracheostomy placement and less likely to have do-not-resuscitate status.
Hanna Vestala, Marcus Bendtsen, Patrik Midlöv
et al.
AbstractObjective To estimate the effects of an interactive web-based support system via mobile phone on preference-based patient participation in patients with hypertension treated in primary care (compared with standard hypertensive care only).Design A parallel group, non-blinded, randomized controlled trial, conducted October 2018–February 2021. Besides standard hypertensive care, the intervention group received eight weeks of support via mobile phone to facilitate self-monitoring and self-management, tentatively providing for augmented patient engagement.Setting 31 primary healthcare centers in Sweden.Subjects 949 patients treated for hypertension.Main outcome measures The effects on preference-based patient participation, that is, the match between a patient’s preferences for and experiences of patient participation in their health and healthcare. This was measured with the 4Ps (Patient Preferences for Patient Participation) tool at baseline, after 8 weeks, and at 12 months. Data were registered electronically and analyzed with multilevel ordinal regression.Results At baseline, 43–51% had a complete match between their preferences for and experiences of patient participation. There was an indication of a positive effect by a higher match for ‘managing treatment myself’ at 8-weeks in the intervention group. Such preference-based participation in their health and healthcare was reversed at 12 months, and no further effects of the intervention on preference-based patient participation persisted after 12 months.Conclusion The interactive web-based support system via mobile phone had a wavering effect on preference-based patient participation. There is a prevailing need to better understand how person-centered patient participation can be facilitated in primary care.
Recently, Large Language Models (LLMs) have showcased remarkable capabilities in natural language understanding. While demonstrating proficiency in everyday conversations and question-answering situations, these models frequently struggle in domains that require precision, such as medical applications, due to their lack of domain-specific knowledge. In this paper, we describe the procedure for building a powerful, open-source language model specifically designed for medicine applications, termed as PMC-LLaMA. Our contributions are threefold: (i) we systematically investigate the process of adapting a general-purpose foundation language model towards medical domain, this involves data-centric knowledge injection through the integration of 4.8M biomedical academic papers and 30K medical textbooks, as well as comprehensive fine-tuning for alignment with domain-specific instructions; (ii) we contribute a large-scale, comprehensive dataset for instruction tuning. This dataset encompasses medical question-answering (QA), rationale for reasoning, and conversational dialogues, comprising a total of 202M tokens; (iii) we conduct thorough ablation studies to demonstrate the effectiveness of each proposed component. While evaluating on various public medical question-answering benchmarks, our lightweight PMCLLaMA, which consists of only 13 billion parameters, exhibits superior performance, even surpassing ChatGPT. All models, codes, datasets can be found in https://github.com/chaoyi-wu/PMC-LLaMA.
Pre-training and fine-tuning have emerged as a promising paradigm across various natural language processing (NLP) tasks. The effectiveness of pretrained large language models (LLM) has witnessed further enhancement, holding potential for applications in the field of medicine, particularly in the context of Traditional Chinese Medicine (TCM). However, the application of these general models to specific domains often yields suboptimal results, primarily due to challenges like lack of domain knowledge, unique objectives, and computational efficiency. Furthermore, their effectiveness in specialized domains, such as Traditional Chinese Medicine, requires comprehensive evaluation. To address the above issues, we propose a novel domain specific TCMDA (TCM Domain Adaptation) approach, efficient pre-training with domain-specific corpus. Specifically, we first construct a large TCM-specific corpus, TCM-Corpus-1B, by identifying domain keywords and retreving from general corpus. Then, our TCMDA leverages the LoRA which freezes the pretrained model's weights and uses rank decomposition matrices to efficiently train specific dense layers for pre-training and fine-tuning, efficiently aligning the model with TCM-related tasks, namely TCM-GPT-7B. We further conducted extensive experiments on two TCM tasks, including TCM examination and TCM diagnosis. TCM-GPT-7B archived the best performance across both datasets, outperforming other models by relative increments of 17% and 12% in accuracy, respectively. To the best of our knowledge, our study represents the pioneering validation of domain adaptation of a large language model with 7 billion parameters in TCM domain. We will release both TCMCorpus-1B and TCM-GPT-7B model once accepted to facilitate interdisciplinary development in TCM and NLP, serving as the foundation for further study.
Interacting agents receive public information at no cost and flexibly acquire private information at a cost proportional to entropy reduction. When a policymaker provides more public information, agents acquire less private information, thus lowering information costs. Does more public information raise or reduce uncertainty faced by agents? Is it beneficial or detrimental to welfare? To address these questions, we examine the impacts of public information on flexible information acquisition in a linear-quadratic-Gaussian game with arbitrary quadratic material welfare. More public information raises uncertainty if and only if the game exhibits strategic complementarity, which can be harmful to welfare. However, when agents acquire a large amount of information, more provision of public information increases welfare through a substantial reduction in the cost of information. We give a necessary and sufficient condition for welfare to increase with public information and identify optimal public information disclosure, which is either full or partial disclosure depending upon the welfare function and the slope of the best response.
Alex P Salam, Alexandre Duvignaud, Marie Jaspard
et al.
Ribavirin is currently the standard of care for treating Lassa fever. However, the human clinical trial data supporting its use suffer from several serious flaws that render the results and conclusions unreliable. We performed a systematic review of available pre-clinical data and human pharmacokinetic data on ribavirin in Lassa. In in-vitro studies, the EC50 of ribavirin ranged from 0.6 μg/ml to 21.72 μg/ml and the EC90 ranged from 1.5 μg/ml to 29 μg/ml. The mean EC50 was 7 μg/ml and the mean EC90 was 15 μg/ml. Human PK data in patients with Lassa fever was sparse and did not allow for estimation of concentration profiles or pharmacokinetic parameters. Pharmacokinetic modelling based on healthy human data suggests that the concentration profiles of current ribavirin regimes only exceed the mean EC50 for less than 20% of the time and the mean EC90 for less than 10% of the time, raising the possibility that the current ribavirin regimens in clinical use are unlikely to reliably achieve serum concentrations required to inhibit Lassa virus replication. The results of this review highlight serious issues with the evidence, which, by today standards, would be unlikely to support the transition of ribavirin from pre-clinical studies to human clinical trials. Additional pre-clinical studies are needed before embarking on expensive and challenging clinical trials of ribavirin in Lassa fever.
Arctic medicine. Tropical medicine, Public aspects of medicine
The world was unprepared for COVID-19. Pandemics can unfold quickly; faster than governments can respond, unless they have maintained a realistic pandemic playbook. As the world ahead becomes ever-more complex, such playbook becomes ever-more necessary. This article not only describes the importance of a pandemic playbook but also a system to maintain it. A pandemic playbook both (1) specifies what is needed to respond to a pandemic and (2) provides a lens through which to identify measures that will keep people safe and society secure. The plays in the book are thought-though policies and strategies and corresponding implementation plans. The process of developing a playbook is as important as the product. Any playbook must be fit for purpose in the context of the times in which it is to be used. Above all, it must contain realistic policies and plans that can actually be implemented and can realize their intended effects. Achieving this goal requires (1) repeatedly exercising the playbook so that people know what to do when they need to do it and (2) evaluating results and updating the playbook to keep it relevant and current. Necessarily, to bring ideas alive, this article illustrates them with reference to COVID-19 and earlier pandemics, but it is not intended as a playbook for responding to the next pandemic; nor a postmortem on responses to COVID-19. Instead, it describes actions to take now to be ready when the next global pandemic strikes, so that policy decision-makers will not be lamenting “we should have done that.”