159370 research outputs found
Sort by
Déjà vu in healthcare AI: lessons from the world’s pioneer AI clinical decision support system
Get PDFAdvancing unpaved road assessment in Africa: Leveraging multimodal machine learning and large language-and-vision assistants across satellite imagery resolutions
No full textOver 53 % of African road network is unpaved, yet systematic monitoring remains limited. This study introduces a cost-effective machine learning (ML) solution to help local authorities monitor and plan road maintenance. Building on earlier work using high-resolution satellite imagery in Tanzania, the analysis extends to Madagascar, incorporating medium- and low-resolution imagery to reduce costs. Two distinct methodologies were evaluated: traditional ML and multimodal ML. The multimodal ML model achieves 93.2 % accuracy with high-resolution imagery and maintains satisfactory performance with medium (84.0 %) and low-resolution (85.3 %) imagery, aided by transfer learning. The framework demonstrates robust cross-resolution performance across Tanzania and Madagascar contexts. Additionally, a pilot study explored a fine-tuned Large Language-and-Vision Assistant (LLaVA) model, which demonstrated potential for natural language-based condition reporting and maintenance recommendations, offering an interpretable alternative to quantitative classification outputs. Whilst LLaVA currently exhibits lower classification accuracy than the multimodal ML model, multi-turn conversational approaches show promise for enhancing performance whilst maintaining natural language interpretability. This study contributes to Sustainable Development Goal 9.1 by delivering a scalable, affordable strategy to support resilient infrastructure and economic development in low-income regions
Landscapes and Cyberscapes of the Commons: Scottish Festivals in the Pandemic and Beyond
Get PDFThis paper investigates how Scottish festivals chose to implement commoning practices in both digital and hybrid spaces during the COVID-19 pandemic. Utilising a deductive thematic analysis approach, the authors draw out emergent themes from a series of interviews with festivals conducted between October 2020 and September 2021 to develop a set of aspects or commonalities between commoning practices for digital and hybrid festivals. The authors examine how some of these commoning practices are directly linked to traditional understandings of the cultural commons, such as the sharing of physical and knowledge resources with other organizations or shared forms of governance. While describing festivals as “commons” in themselves would be reductive and conceptually contentious, this paper argues not only that festivals can include commoning practices, but that these are inherently relevant to their identity and audiences that, even in times of extreme operational restrictions, these remain central to their activities
Cross-sectional illusions: what we have learned about the attitude-behaviour relationship and its policy implications
Get PDFWe describe and challenge long-standing assumptions in transport research about the direction and strength of the relationship between attitudes and behaviour. Economic and social science theories suggest a one-way effect from attitudes (interchangeably perceptions or motives) to behaviour. Drawing on a synthesis of empirical studies focused on car use and ownership, we show that this view is simplistic. Most research tests this relationship through cross-sectional data and reports medium to large effects from attitude to behaviour (behavioural intention). However, in modern (travel) behavioural modelling, emerging (longitudinal) panel models reveal that: (i) the attitude-behaviour relationship is bidirectional, (ii) the strength of the real effects is weaker than what is suggested by cross-sectional studies, (iii) attitudes are more a function of behaviours, not the other way around, and (iv) behaviours are more a function of past behaviours than of deliberate planning; contrary to the assumptions of the theory of planned behaviour. From a policy perspective, expecting to change (travel) behaviour solely by changing attitudes, often referred to as soft or pull measures, may be overly optimistic. We advise researchers to be cautious when using cross-sectional data to inform policy decisions. Directions for future research are also discussed
Knockdown of the fly spliceosome component Rbp1 (orthologue of SRSF1) extends lifespan
Get PDFBiological regulation is an intricate process involving many layers of complexity, including at the RNA level. Alternative splicing is crucial in the regulation of which components of a protein-coding gene are spliced into a translatable mRNA. During ageing, splicing becomes dysregulated, and alternative splicing is implicated in disease and known anti-ageing treatments such as dietary restriction (DR) and mTOR suppression. In prior work, we have shown that DR and mTOR suppression modulate the expression of the spliceosome in the fly (Drosophila melanogaster). Here, we manipulated the five top genes that change in expression in both these treatments. We found that knockdown (using conditional in vivo RNAi in adults) of some spliceosome components rapidly induces mortality, whereas one, Rbp1, extends lifespan. Treatments that have more instant benefits on longevity are more translatable. We therefore subsequently repeated the Rbp1 experiment but initiated Rbp1 knockdown at later stages in adult life. We find that irrespective of the age of induction, knockdown of Rbp1 extends lifespan. Our results posit the spliceosome itself as a hub of regulation that when targeted can extend lifespan, rendering it a promising target for geroscience
Transforming jet flavour tagging at ATLAS
Get PDFJet flavour tagging enables the identification of jets originating from heavy-flavour quarks in proton–proton collisions at the Large Hadron Collider, playing a critical role in its physics programmes. This paper presents GN2, a transformer-based flavour tagging algorithm deployed by the ATLAS Collaboration that represents a different methodology compared to previous approaches. Designed to classify jets based on the flavour of their constituent particles, GN2 processes low-level tracking information in an end-to-end architecture and incorporates physics-informed auxiliary training objectives to enhance both interpretability and performance. Its performance is validated in both simulation and collision data. The measured c-jet (light-jet) rejection in data is improved by a factor of 3.5 (1.8) for a 70% b-jet tagging efficiency, compared to the previous algorithm. GN2 provides substantial benefits for physics analyses involving heavy-flavour jets, such as measurements of Higgs boson pair production and the couplings of bottom and charm quarks to the Higgs boson, and demonstrates the impact of advanced machine learning methods in experimental particle physics
Investigations of Base Simulation CAD Packages for Initial Regulatory Feasibility Testing in Medical SMEs
Get PDFThe development of medical devices is a complex and highly regulated
process, often requiring skilled engineers to conduct advanced analyses for
mechanical safety and efficacy. However, med-tech SMEs with limited resources
face challenges in navigating these requirements. Industrial designers play a crucial
role in making devices both functional and competitive by incorporating
ergonomics, aesthetics, and user comfort, and recently basic FEA analysis. To
support this, we propose a framework showing how Industrial Designers can utilize
basic linear and static simulation tools, allowing for iterative design refinements
that align with regulatory standards before significant investment
Understanding the roles of decommodification in socioecological transformations: a new theoretical approach
Get PDFDecommodification is widely considered a central pillar of progressive socioecological transformations, like the green new deal, degrowth and ecosocialism. Yet it is inadequately problematised and theorised in the existing literature. It is often simply stated as a normative goal rather than theorised as a central part of the transformative process. It is also commonly reduced to services, such as transport and energy, and sometimes framed in Eurocentric terms. The global geographical diversity of decommodification is therefore overlooked and the wider role it plays in both supporting and challenging capitalism is underexplored. This article introduces a new theoretical approach that captures these dimensions and provides a stronger foundation to analyse and strategise the diverse roles of decommodification in socioecological transformations. It argues that the relationship between decommodification, production, and value is crucial for grasping the character of decommodification and its capacity to support or hinder progressive socioecological transformations. Further, the article shows that although decommodification is central to progressive transformations, it can also support regressive and reactionary responses to environmental crises. Hence, simply calling for more decommodification is not enough; grasping its limits and contradictions, and tailoring it to the context and task are essential
Data integrity in materials science in the era of AI: balancing accelerated discovery with responsible science and innovation
Get PDFArtificial intelligence promises to revolutionise materials discovery through accelerated prediction and optimisation, yet this transformation brings critical data integrity challenges that threaten the scientific record. Recent studies demonstrate that experts cannot reliably distinguish AI-generated microscopy images from authentic experimental data, while widespread errors plague 20–30% of materials characterisation analyses. Generative AI tools can now produce code for data manipulation at pace, creating plausible-looking results that violate fundamental physical principles yet evade traditional peer review. These risks are compounded by inherent biases in training datasets that systematically over represent equilibrium-phase oxide systems, and by the “black box” opacity of AI models that challenges scientific accountability and epistemic agency. We propose a multifaceted framework for enhanced research integrity encompassing materials-specific ethical governance, professional standards for AI disclosure and data validation, and modular integrity checklists with technique-specific validation protocols. Critical enablers include mandatory deposition of structured raw instrument files, AI-powered fraud detection systems, and cultivation of critical AI literacy through interdisciplinary education. Without immediate action to address these challenges, the materials science community risks perpetuating errors and biases that will fundamentally undermine AI's transformative potential
Flow‐pattern evolution of the Scandinavian Ice Sheet indicated by the subglacial lineation record over Norway, Sweden and Finland
Get PDFStreamlined subglacial bedforms, including drumlins, mega‐scale glacial lineations, crag‐and‐tails and roche moutonnées, provide evidence for the past flow of large mid‐latitude ice sheets during the late Quaternary. Empirical reconstructions of palaeo‐ice sheet flow, based on such landforms, provide valuable insights into how ice sheets evolve over time and adjust their internal dynamics in response to climate change. We present a new 25‐stage reconstruction of changing flow directions of the Scandinavian Ice Sheet (SIS) based on systematic mapping of ~240 000 subglacial bedforms across Norway, Sweden, Finland and parts of NW Russia. Of these, 23 stages depict the ice flow evolution during advance and retreat of the SIS through Marine Isotope Stage (MIS) 2. Two additional stages likely represent flow patterns of an earlier ice sheet (potentially MIS 4 or older). Our reconstruction was enabled by the recent revolution in the availability of high‐resolution (1–2 m) digital terrain models. It is based on 611 flowsets, which summarise discrete ice‐flow patterns recorded by subglacial lineations and are individually categorised by their glaciodynamic contexts. The reconstruction honours the relative‐age chronology of flowsets indicated by cross‐cutting relationships of the subglacial lineations. We reconstruct, and provide maps of, changing ice‐sheet flow patterns and the migration of ice divides starting with ice‐sheet inception, through advance and subsequent deglaciation, and ultimately the fragmentation into independent ice masses. The primary ice divide migrated up to 500 km and developed a branched configuration during deglaciation. The reconstruction of SIS flow patterns we present is the most detailed and comprehensive to date, and the fact that we independently verify many properties of the ice sheet invoked by earlier workers is testament to the quality, rigour and enduring legacies of those studies. We release flowsets, relative chronology and flow‐pattern data along with a dataset of ~58 000 lineation linkages which summarise our detailed landform mapping and were invaluable for reconstructing ice‐flow patterns at the ice‐sheet‐scale. In releasing these data, we intend for them to serve as useful inputs or comparative data for future studies in palaeoglaciology. This includes, for example, approaches combining flow pattern information with numerical ice sheet modelling to improve representations of ice sheet behaviour. Such improvements should yield increased robustness of information on time‐varying glacio‐isostatic loading by the ice sheet, relevant for sea‐level forecasting. Our datasets also have wide utility for applications beyond palaeoglaciology, such as for mineral exploration
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇