179566 research outputs found
Sort by
A self-supervised learning framework for soft robot proprioception
Get PDFThe inherent compliant nature of soft robots can offer remarkable advantages over their rigid counterparts in terms of safety to human users and adaptability in unstructured environments. However, this feature also magnifies the complexityof their bodies, rendering their proprioception, and hence their control, extremely challenging. Given this intricacy, machine learning is a potent candidate for extracting proprioceptive insights from sensor data due to its proven capabilities in tackling analogous issues in computer vision and natural language processing.Recently, key aspects of soft robot proprioception have been addressed via learning-based techniques, but most of these are rooted in the supervised learning paradigm. This typically requires collecting a large number of costly annotated training samples, thereby constraining its widespread and speedy adoption in real-world applications. To mitigate this limitation, we proposea self-supervised learning framework for soft robot proprioception. Our method utilizes vast unannotated data for network pretraining by self-supervised learning. Then, the pretrained model is finetuned with a limited set of annotated samples by supervised learning. We validate the proposed method’s efficacy on a high-resolution 3D morphological reconstruction task using a publicly available dataset. Remarkably, our approach is shown to necessitate only about 1/20 of annotated samples to achieve better performance than the fully supervised method
Injury reporting and the use of injury prevention programmes in women’s compared with men’s rugby union players:A scoping review
Get PDFObjectiveIdentify current injury surveillance and prevention literature in women’s compared to men’s rugby union players.DesignScoping review.MethodsA two-step search strategy identified relevant published and unpublished literature on adult rugby players from five electronic databases, three governing body season report storage locations, and citation searches. Literature was screened against inclusion criteria for time-loss injury and/or injury prevention programmes and outcomes. Data was extracted and findings were reported using (1) a numerical analysis and (2) a thematic analysis.ResultsIn total, 3196 articles were screened at abstract and full-text level, 252 met inclusion criteria. Across 252 studies were 330 cohorts, women-only cohorts accounted for 24% (n = 79) of injury surveillance literature. Match injury incidence ranges were greater than training across all cohorts, men’s and women’s injury rates and severity across match and training were similar. Only 30% of cohorts assessed training injury, 27% in men’s and 42% in women’s cohorts. General agreement highlighted lower limb, joint/ligament and concussion injuries to be most common across the men’s and women’s game. Risk factors were commonly assessed in men’s literature (49%) but reported less within women’s research (25%). Thirteen injury prevention studies were identified, two involved evaluation of injury prevention initiatives in women’s cohorts.ConclusionThere is limited women’s representation in rugby injury surveillance research compared to men’s, and there is scarce evidence of the implementation and evaluation of injury prevention initiatives to reduce injury rates in women. Future research should focus on women’s surveillance to inform injury prevention studies, implemented and evaluated in women’s rugby cohorts
TSPRank:Bridging pairwise and listwise methods with a bilinear travelling salesman model
Get PDFTraditional Learning-To-Rank (LETOR) approaches, including pairwise methods like RankNet and LambdaMART, often fall short by solely focusing on pairwise comparisons, leading to sub-optimal global rankings. Conversely, deep learning based listwise methods, while aiming to optimise entire lists, require complex tuning and yield only marginal improvements over robust pairwise models. To overcome these limitations, we introduce Travelling Salesman Problem Rank (TSPRank), a hybrid pairwise-listwise ranking method. TSPRank reframes the ranking problem as a Travelling Salesman Problem (TSP), a well-known combinatorial optimisation challenge that has been extensively studied for its numerous solution algorithms and applications. This approach enables the modelling of pairwise relationships and leverages combinatorial optimisation to determine the listwise ranking. TSPRank can be directly integrated as an additional component into embeddings generated by existing backbone models to enhance ranking performance. Our extensive experiments across three backbone models on diverse tasks, including stock ranking, information retrieval, and historical events ordering, demonstrate that TSPRank significantly outperforms both pure pairwise and listwise methods. Our qualitative analysis reveals that TSPRank's main advantage over existing methods is its ability to harness global information better while ranking. TSPRank's robustness and superior performance across different domains highlight its potential as a versatile and effective LETOR solution
Observed Seasonality of Mixed-Layer Eddies and Vertical Heat Transport Over the Antarctic Continental Shelf
Get PDFThe cruel optimism of suicide prevention:Thinking beyond the mental health model
Get PDFSuicide has long been constructed as an individual, pathological problem of the mind, requiring a combination of clinical care and interpersonal support to prevent it. The endurance of this individualising and pathologising approach serves to reflect and maintain the marginalisation of sociology in suicide studies. This article contributes to redressing this balance, via analysis of creative, qualitative workshops which formed part of a broader study exploring the politics of suicide. Informed by 33 participants’ contributions from six creative-response workshop groups, in dialogue with Lauren Berlant’s concept of ‘cruel optimism’, we propose a creative sociological (re)turn in suicide studies. Our analysis explores how the politics of UK suicide prevention constitutes a form of ‘cruel optimism’, obfuscating structural and sociological approaches to preventing suicide, and giving primacy to individual mental health-led interventions. We argue this provides a vital opportunity and urgent call for sociology to contribute more robustly to suicide research
All the vista we cannot see: Scenes of wandering in French migrant psychiatry
Get PDFThese five prose poems engage French migrant psychiatry as an ethnographic object in the scene of wandering. They draw from observations of psychiatric consultations and scenes of migrant life in Paris, following ways homelessness and asylum refusal move through migrant life and clinical encounters. Poem (I) reflects on a clinical interview with a mother separated from her son at the border with Turkey, her protective silence; (II) reprises the colonial present in a searing scene of four lifeless bodies on a traffic junction; in (III), the case of a homeless man nursing a brain injury highlights displacements in the shifts of psychiatric practice between idealism, pragmatism, and complicity with the political order; (IV) exposes gendered power in the therapeutic relation, in the texture of chase and negotiation; (V) details a psychiatrist's work to craft the seeds of hope for a young woman from Congo who has witnessed the killings of all her relatives. The poems question ethnographic knowing in the site of displacement, if knowing might emerge more humanely from the changed relation of an oblique view, and exceed the routing of migrant alterity within political and clinical orthodoxy. This exposes a vista of all we cannot see
Efficacy and Safety of Avutometinib ± Defactinib in Recurrent Low-Grade Serous Ovarian Cancer: Primary Analysis of ENGOT-OV60/GOG-3052/RAMP 201
Get PDFPURPOSE This study evaluated the efficacy and safety of avutometinib (rapidly acceleratedfibrosarcoma/mitogen-activated extracellular signal-regulated kinase [MEK]clamp) alone or in combination with defactinib (focal adhesion kinase inhibitor)in patients with recurrent low-grade serous ovarian cancer (LGSOC).METHODS In this phase II, open-label study, patients with recurrent, measurable LGSOCafter ≥1 line of platinum chemotherapy were stratified by tumor Kirsten ratsarcoma virus homolog (KRAS) mutation status and randomly assigned to oralavutometinib 4.0 mg two times per week monotherapy or avutometinib 3.2 mgtwo times per week in combination with oral defactinib 200 mg two times perday. The combination was selected as the go-forward regimen for expansion.The primary end point was objective response rate (ORR) by blinded independentcentral review.RESULTS A total of 115 patients received the go-forward combination regimen. Patientshad a median of 3 (range, 1-9) prior lines of therapy, including hormonal (86%),bevacizumab (51%), and MEK inhibitor (22%). Confirmed ORR was 31% (95%CI, 23% to 41%) with a median duration of response of 31.1 months (95% CI,14.8 to 31.1). ORR was 44% in KRAS-mutant and 17% in KRAS wild-type cohorts.The median progression-free survival was 12.9 months (95% CI, 10.9 to 20.2)overall and 22.0 months (95% CI, 11.1 to 36.6) and 12.8 months (95% CI, 7.4 to18.4) in KRAS-mutant and wild-type cohorts, respectively. The most frequentgrade ≥3 treatment-related adverse events (AEs) were elevated creatinephosphokinase (24%), diarrhea (8%), and anemia (5%). Ten percent of patientsdiscontinued because of AEs.CONCLUSION The efficacy and safety profile of avutometinib in combination with defactinibsupport this combination as a potential standard of care for recurrent LGSOC. Arandomized phase 3 study of avutometinib and defactinib versus investigator’schoice of therapy for women with recurrent LGSOC is currently enrolling(RAMP301; ClinicalTrials.gov identifier: NCT06072781)
Towards explainable graph embeddings for gait assessment using per-cluster dimensional weighting
Get PDFAs gait pathology assessment systems improve both in accuracy and efficiency, the prospect of using these systems in real healthcare applications is becoming more realistic. Although gait analysis systems have proven capable of detecting gait abnormalities in supervised tasks in laboratories and clinics, there is comparatively little investigation into making such systems explainable to healthcare professionals who would use gait analysis in practice in home-based settings. There is a “black box” problem with existing machine learning models, where healthcare professionals are expected to “trust” the model making diagnoses without understanding its underlying reasoning. To address this applicational barrier, an end-to-end pipeline is introduced here for creating graph feature embeddings, generated using a bespoke Spatio-temporal Graph Convolutional Network and per-joint Principal Component Analysis. The latent graph embeddings produced by this framework led to a novel semi-supervised weighting function which quantifies and ranks the most important joint features, which are used to provide a description for each pathology. Using these embeddings with a K-means clustering approach, the proposed method also outperforms the state-of-the-art by between 4.53% - 16% in classification accuracy across 3 datasets with at total of 14 different simulated gait pathologies from minor limping to Ataxic gait. The resulting system provides a workable improvement to at-home gait assessment applications by providing accurate and explainable descriptions of the nature of detected gait abnormalities without need of prior labeled descriptions of detected pathologies.<br/
Encoding Product Types
Get PDFCan product types be encoded in simply-typed lambda calculus with base types and function types? It depends
Greater recovery after critical illness (GRACE): a call to action to create a new roadmap for critical illness research
Get PDFFor decades, most critical care patients have survived hospitalisation, supporting increased attention on the long-term critical illness recovery. The term ‘Post-Intensive Care Syndrome’ was coined in 2012 to raise awareness of long-term impairment in physical, cognitive and/or mental health after critical illness. However, the incidence of these impairments has persisted over the past decade, reaching as high as 60% and remains a major public health problem.Aiming to set a research agenda to address evidence gaps in critical illness recovery over the next 10 years, we invited key international opinion leaders from diverse clinical and methodological backgrounds to a roundtable meeting in June 2024 to assess the progress of post-critical illness recovery research and outline a future research agenda to address the unmet needs of critical illness survivors over the next decade.An early outcome from the meeting was to conduct a thematic analysis of critical care recovery literature, which highlighted the need for effective expectation management, ongoing patient support and education throughout recovery, integration between inpatient and community care, caregiver support and opportunities to reconnect with the intensive care unit.Participants identified conceptual challenges concerning current terminology and scope, population heterogeneity and phenotyping, and outcome definitions. Methodological challenges were identified around study design, with a call to shift to contemporary trial designs, incorporating qualitative methods. Translation into clinical practice will require interdisciplinary engagement.The roundtable concluded that a roadmap should be developed to guide clinical and research efforts over the coming decade, with the aim of developing a precision recovery approach
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!