179566 research outputs found
Sort by
Differential benefit of adjuvant everolimus according to endocrine therapy backbone in the randomized UNIRAD trial
No full textBackgroundThe randomized, double-blind UNIRAD trial evaluating the addition of 2 years of everolimus to endocrine therapy in patients with high-risk, early luminal breast cancer failed to demonstrate a benefit. We report the subgroup analyses.Patients and methodsWe randomly assigned 1278 patients in a 1 : 1 ratio to receive 2 years of placebo or everolimus, added to endocrine therapy for up to 4 years after initiation. Randomization was stratified by endocrine therapy agent, prior adjuvant versus neoadjuvant therapy, progesterone receptor expression, and lymph node involvement. Subgroup analyses by each stratification factor were pre-specified. Post hoc analyses were carried out according to menopausal status and age. Treatment adherence was also analyzed.ResultsWe observed a limited trend toward more favorable prognostic features in tamoxifen-treated patients, with more frequent estrogen receptor-positive/progesterone receptor-positive tumors (88.5% versus 84.1%, P = 0.026) and less frequent pN2-positive status (39.8% versus 46.0%, P = 0.032). In premenopausal women, we observed a numerical benefit of everolimus: 3-year disease-free survival was 86% in the placebo group and 90% in the everolimus group (hazard ratio 0.76, 95% confidence interval 0.43-1.34). In premenopausal patients treated with tamoxifen (n = 153; 12.3%), we observed an even stronger trend in favor of everolimus as 3-year DFS was 84% in the placebo group and 91% in the everolimus group (hazard ratio 0.54, 95% confidence interval 0.28-1.02). Early discontinuation of either everolimus or placebo was less frequent in the tamoxifen group than in the aromatase inhibitor group: 48.0% versus 56.9% (P = 0.028).ConclusionsThe present post hoc analyses generate hypotheses regarding the interaction between menopausal status, tamoxifen, and everolimus in patients with high-risk, ER-positive, human epidermal growth factor receptor type 2-negative early breast cancer. They suggest that tamoxifen alone is an underpowered endocrine treatment in high-risk premenopausal patients
Volatomics for Diagnosis and Risk Stratification of MASLD: A Proof-of-Concept Study
No full textBackground & AimsHuman breath contains numerous volatile organic compounds (VOCs) produced by physiological and metabolic processes or perturbed in pathological states. Electronic nose (eNose) technology has been extensively validated as a non-invasive diagnostic tool for respiratory disease. Using eNose-derived exhaled breath signals, we investigated whether it could discriminate patients with metabolic dysfunction-associated steatotic liver disease (MASLD) from healthy volunteers and identify patients at high risk of disease progression. MethodsIn a prospective single-centre study, exhaled breath VOCs were analysed using an eNose, in a well-characterized cohort comprising patients with Child-Turcotte-Pugh class A MASLD cirrhosis (n=30), non-cirrhotic MASLD (n=30) and healthy volunteers (n=30). An unbiased machine learning clustering technique was applied. Longitudinal clinical data were collected over five years for patient cohort. Logistic regression and univariable analysis were performed to identify risk factors for disease progression, liver-related outcomes, and all-cause mortality. ClinicalTrials.gov identifier: NCT02950610.ResultsPrincipal component analysis of breath VOCs discriminated patients with MASLD from healthy volunteers with 100% sensitivity (p<0.001, cross-validation verification of 96%), independent of age and gender. The eNose breath profile classified patients with MASLD into three distinct subgroups with similar baseline clinical and demographic characteristics but markedly different prognoses. During the 5-year follow-up period, Cluster 2 was identified as a higher-risk subgroup for progression (42%, p=0.03), liver-related decompensation events (17%, p=0.06), and all-cause mortality (12.5%).ConclusioneNose can discriminate patients with MASLD from healthy volunteers and, using unbiased clustering analysis, identify patients with a significantly worse prognosis. These results warrant prospective validation in independent MASLD populations.<br/
Comparative mutant analyses reveal a novel mechanism of ARF regulation in land plants
No full textThe plant hormone auxin regulates a wide variety of transcriptional responses depending on the cell type, environment and species. How this diversity is achieved may be related to the specific complement of auxin-signalling components in each cell. The levels of activators (class-A AUXIN RESPONSE FACTORS) and repressors (class-B ARFs) are particularly important. Tight regulation of ARF protein levels is probably key in determining this balance. Through comparative analysis of novel, dominant mutants in maize and the moss Physcomitrium patens, we have discovered a ~500-million-year-old mechanism of class-B ARF protein-level regulation mediated by proteasome degradation, important in determining cell fate decisions across land plants. Thus, our results add a key piece to the puzzle of how auxin regulates plant development
Open-Amp:Synthetic data framework for audio effect foundation models
No full textThis paper introduces Open-Amp, a synthetic data framework for generating large-scale and diverse audio effects data. Audio effects are relevant to many musical audio processing and Music Information Retrieval (MIR) tasks, such as modelling of analog audio effects, automatic mixing, tone matching and transcription. Existing audio effects datasets are limited in scope, usually including relatively few audio effects processors and a limited amount of input audio signals. Our proposed framework overcomes these issues, by crowdsourcing neural network emulations of guitar amplifiers and effects, created by users of open-source audio effects emulation software. This allows users of Open-Amp complete control over the input signals to be processed by the effects models, as well as providing high-quality emulations of hundreds of devices. Open-Amp can render audio online during training, allowing great flexibility in data augmentation. Our experiments show that using Open-Amp to train a guitar effects encoder achieves new state-of-the-art results on multiple guitar effects classification tasks. Furthermore, we train a one-to-many guitar effects model using Open-Amp, and use it to emulate unseen analog effects via manipulation of its learned latent space, indicating transferability to analog guitar effects data
Artificial Intelligence-driven prediction of optimal technology-aided alternative operations in post-emergency contexts:A case study from an Emirati University
No full textDespite the challenges posed by the recent pandemic, educational institutions were prompted to explore alternative operation modes to enhance teaching, learning, and service delivery through technology. However, effective implementation of these technology-aided modes in post-emergency contexts necessitates evidence-based practices and contextual insights into stakeholders’ challenges, comfortability, and preferences. This study aims to support the efficient planning and execution of digital transformation within a university in the United Arab Emirates (UAE) by examining stakeholders’ comfortability, challenges, and preferences following the pandemic’s impact. The novelty of this research lies in its use of artificial intelligence, specifically fuzzy logic, to predict stakeholder preferences, complemented by comprehensive stakeholder-centric analysis and an in-depth examination of demographic influences on digital transformation preferences. Additionally, the study provides unique regional insights within the UAE context, addressing cultural, economic, and technological factors underrepresented in international literature. Utilizing a survey method, data were analyzed through descriptive statistics and AI-driven predictive analytics. Findings indicate that institutional support and familiarity with online platforms reduced stress during the transition to technology-aided modes, with a strong preference for hybrid flexible models influenced significantly by demographic factors. This study contributes by demonstrating the enhanced predictive capabilities of AI in understanding stakeholder needs, offering tailored digital transformation strategies, highlighting the importance of demographic considerations, and providing a practical roadmap for building a sustainable and resilient digital ecosystem. Furthermore, it informs educational policy and governance, ensuring that technology-aided operations are effectively planned and implemented to meet the evolving needs of the academic community in post-emergency settings
Adaptive Activation Steering: A Tuning-Free LLM Truthfulness Improvement Method for Diverse Hallucinations Categories
No full textRecent studies have indicated that Large Language Models (LLMs) harbor an inherent understanding of truthfulness, yet often fail to consistently express it and generate false statements. This gap between ''knowing'' and ''telling'' poses a challenge for ensuring the truthfulness of generated content. Inspired by recent work on the practice of encoding human-interpretable concepts linearly within large language models, we treat truthfulness as a specially linearly encoded concept within LLMs, and introduce Adaptive Activation Steering (ACT), a tuning-free method that adaptively shifts LLM's activations in the ''truthful'' direction during inference. ACT addresses diverse categories of hallucinations by utilizing diverse truthfulness-related steering vectors and adjusting the steering intensity adaptively. Applied as an add-on across various models, ACT significantly improves truthfulness in LLaMA (↑142%), LLaMA2 (↑24%), Alpaca (↑36%), Vicuna (↑28%), LLaMA2-Chat (↑19%), and LLaMA3(↑34%). Furthermore, we verify ACT's scalability across larger models (13B, 33B, 65B), underscoring the adaptability of ACT to large-scale language models. Our code is available at https://github.com/tianlwang/ACT.<br/
A comparison of generic and subject-specific finite element models of distal femur fractures treated with locking plates
No full textWhile the need for employing subject-specific computational biomechanics models for treatment planning in orthopaedics is being increasingly voiced it has not been clear when such specificity is essential and for which questions simpler models might be adequate. This study uses a novel modelling approach to generate finite element models to examine the influence of subject-specificity in the treatment of distal femur fractures. Three subject-specific finite element models are created from clinical CT scans and the proposed approach employed to impose identical fractures and locking plate treatments upon them. Additionally, the performance of the generic two-material model based on a Sawbones fourth generation femur is also evaluated. Interfragmentary motions, plates stresses, and strains at the screw-bone interface are examined due to a physiological loading at different stages of healing. The study finds that subject-specificity has a major effect on strains in the bone at the screw-bone interface. However, interfragmentary motions at the far cortex and plate stresses show minimal sensitivity to subject-specific factors, while near-cortical and shear interfragmentary motions are influenced by them. The influence of subject-specificity decreases as healing progresses. These results indicate that while generic approaches may be sufficient to calculate global assembly responses, material heterogeneity and subject-specific bone stock variations have a large impact on the interaction between the screws and bone. The study also shows that the proposed method which enables manipulating bone geometry while retaining subject-specific properties can be used to evaluate the influence of subject-specificity for other orthopaedic simulations
Digital drug trading ecologies in context:Technological, geographic, and linguistic variation across darknet platforms
No full textBackgroundPrevious research on darknet drug markets has primarily concentrated on large, English-language cryptomarkets, often overlooking regionally oriented platforms that operate in national languages. This study adopts a comparative, exploratory approach to examine how drug trade practices vary across linguistic, geographic, and technological contexts. We introduce the concept of “drug trading ecologies” to describe how platform features, communication norms, and localized settings together shape distinct trading environments.MethodsUsing a mixed-methods approach, we analyzed web-crawled data from three Tor-based platforms: Tsatti (Finnish-language chat), Cebulka (Polish-language forum), and Nemesis (English-language cryptomarket). Data were collected through customized web scraping and analyzed using statistical tools and qualitative content coding to examine platform-specific patterns. Our comparative approach highlights structural and localization-specific variations without attempting exhaustive conceptual definitions.ResultsEach platform displayed a distinct configuration shaped by its technical affordances and localization-specific user practices. Tsatti supported fast, hyperlocal, and highly anonymized exchanges with minimal user identity or community features. Cebulka enabled semi-public vendor-buyer interactions, trust-building through discourse, and diverse product bundling. Nemesis functioned as a transnational, professionalized cryptomarket with standardized listings, formalized trust mechanisms, and branding strategies.ConclusionsRather than attributing differences solely to local, transnational, or design factors in isolation, we argue that darknet drug trading ecologies emerge from the intersection of platform architecture, localization (geographic scope), and language. Our findings underscore the importance of considering these factors when conducting digital ethnography in illicit economies and providing concrete entry points for tailoring harm reduction interventions responsive to the diverse realities of online drug trading.Keywords: Darknet markets; Drug trade ecologies; Digital platforms; Online drug markets; Cryptomarkets; Virtual communities; Illicit economies.<br/
Assessing ancient conflict landscapes through KOCOA analysis:The case of Burnswark hillfort (SW Scotland)
Get PDFThis paper applies KOCOA terrain analysis to the study of the Iron Age hillfort of Burnswark Hill (SW Scotland) and its associated Roman military remains. The Roman camps and projectiles identified at Burnswark have sparked a long scholarly debate, with views ranging from authors that interpret the evidence as related to Roman military training at an already abandoned hillfort, and others who consider it representative of a brutal Roman military attack on an indigenous stronghold. This article presents the recently undertaken KOCOA analysis of the site, which assesses how the terrain influenced the conduct of the Roman intervention at Burnswark. The results are used in conjunction with the insights provided by the metal detector surveys and excavations from the last decade. The combined evidence strongly suggests that the events at Burnswark can best be described as a Roman oppugnatio longinqua obsidio, an active siege that ended in a storming assault.</p
Regulatory T cell depletion promotes myeloid cell activation and glioblastoma response to anti-PD1 and tumor-targeting antibodies
Get PDFGlioblastoma is invariably lethal and responds poorly to immune checkpoint blockade. Here, we examined the impact of regulatory T (Treg) cell depletion on glioblastoma progression and immunotherapy responsiveness. In human glioblastoma, elevated Treg cell signatures correlated with poorer survival outcomes, with these cells expressing high levels of CD25. In Nf1-/-Pten-/-EGFRvIII+ glioblastoma-bearing mice, a single dose of non-interleukin-2 (IL-2) blocking (NIB) anti-CD25 (anti-CD25NIB) antibody depleted Treg cells and promoted CD8+ T cell clonal expansion and partial tumor control, further enhanced by programmed cell death-1 (PD1)-blockade. Treg cell depletion induced interferon-γ (IFN-γ)-dependent tumor microenvironment remodeling, increasing Fcγ receptor (FcγR) expression on intratumoral myeloid cells and enhancing phagocytosis. Combination of anti-CD25NIB with anti-EGFRvIII tumor-targeting antibodies resulted in complete tumor control. Anti-human CD25NIB treatment of glioblastoma patient-derived tumor fragments effectively depleted Treg cells and activated CD8+ T cells. These findings underscore the therapeutic relevance of Treg targeting in glioblastoma and unveil potent combination strategies for anti-CD25NIB based on innate cell activation.</p
Access Repository Dashboard
Do you manage Edinburgh Research Explorer? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!