260851 research outputs found
Sort by
Understanding the role of Anatomy Demonstrators in the United Kingdom
No full textThe employment of Anatomy Demonstrators (ADs) has grown unexpectedly over two decades to meet increasing educational demands. However, their role is poorly understood and defined. There is therefore a need to better understand ADs, to maximise their potential as educators and build the integrity and transparency of UK anatomical education. Through a mixed methods approach this study explored ADs: demographics, role descriptors, experiences within the role, barriers to the role, and the potential of ADs. An online questionnaire (n=50) and three virtual focus groups (N=13) were analysed using thematic analysis to identify emerging patterns and themes. The findings highlight that ADs are generally young people from diverse backgrounds with an even gender split. The role and its responsibilities seemingly vary substantially between universities, with ADs employed in numerous different ways. Being an AD appears to be an overall positive experience and ADs enjoy their work. Furthermore, ADs care about anatomical education and want to engage in discussions to improve their role and develop medical education. This study has expanded the understanding of the UK AD role and concludes that the UK AD role is a varied and often rewarding post that is enjoyed by diverse, engaged individuals who are primarily resident doctors by background. This study provides practical recommendations for anatomy departments to improve the experience for ADs and hopes to stimulate greater interest in ADs in order to maximise their educational potential.</p
Assisting annotators of wearable activity recognition datasets through automated sensor-based suggestions
No full textThis dissertation focuses on exploring and evaluating the potential of methods for annotation assistance in human activity recognition. Annotating sensor signals for activity recognition can be a long, tedious, and repetitive task for annotators. I postulate that an automated annotation assistance system, which draws the attention of annotators to suggested change points and proposes labels for activities, can help speeding up the annotation process and reduce the mental workload without compromising annotation quality. Automated assistance refers to machine learning based methods to support an annotator who performs data annotation tasks identifying physical activities in sensor data recorded by wearable sensors. Suitable methods to assist annotation were selected based on their performance and implemented in an annotation tool. A new web-based annotation tool named Smart Annotation Assistance Tool (SAAT) was developed, which allows users to annotate activities from sensor signals which may be synchronised with a video stream. The assistance is implemented in the form of segmentation suggestions by indicating potential change points of activities and slowing down playback when approaching them. When selecting an annotation at a selected point, the user is offered classification assistance in the form of label suggestions.Different methods were utilised to suggest change points and annotation labels: Hand-crafted suggestions with introduced errors and noise were used to create a controlled environment in which the impact of assistance methods can be isolated. Attention-based models were deployed to improve prediction capabilities by utilising spatial and temporal relationships within the sensor signals. In addition to supervised methods, semi-supervised machine learning methods were investigated to utilise the potential of unlabelled data. State-of-the-art generative adversarial networks were deployed to augment training data with synthetic samples.The suggested change points and annotations were introduced into the assisted annotation tool. The quantitative and qualitative impact on time, stress, and annotation performance of these forms of assistance was evaluated in two exhaustive user studies. The results show that the implemented annotation assistance improved the annotation quality by 11% F 1 Score but reduced annotation speed by 20%, whereas the feedback on the perceived mental workload show that participants experienced the assistance as beneficial for annotation speed but not for annotation quality.This perception is in line with our expectations prior to carrying out the experiment and the miss-match with the measured results highlights interesting areas for future research. Those include investigating measures to correct users perceptions of the impact of annotation assistance.</p
Recognising the needs of queer Iranian migrants
No full textNew research shows that queer Iranian migrants require enhanced logistical, legal and medical support in host countries, and faster asylum processing by SOGIESC-aware staff.</p
Navigating competing priorities for societal value creation: tensions in sustainability-driven enterprises across venture stages
No full textIn the pursuit of impact, entrepreneurs encounter various sustainability tensions throughout the development stages of their venture. Recognizing these tensions is critical for navigating and negotiating competing priorities in societal value creation. However, existing studies on sustainability tensions have mainly focused on large and incumbent firms. While studies exploring tensions in small enterprises have started to emerge, limited attention has been given to sustainability-driven enterprises and the types of tensions they encounter as they move through different venture stages in the pursuit of and scaling for impact. In this paper, we aim to fill this research gap by examining the tensions that sustainability-driven enterprises experience at different venture stages. We employ a multiple case study approach to identify tensions that entrepreneurs encounter in pursuing social and environmental impacts. Our findings identify tensions at three levels—individual, organizational, and macro-level context—and highlight how different tensions dominate at different stages, each carrying distinct implications for impact creation and scaling.</p
The Battle of the Bonus Army
No full textWhen angry US army veterans marched on Washington during the Great Depression, their protest exposed deep fault lines between military loyalty and political authority.</p
Experience of responding to imaginative suggestions: a micro-phenomenological interview exploratory study
No full textThe micro-phenomenological interview is a technique for invesNgaNng phenomenology in detail during an acNvity lasNng seconds to minutes. Here, we applied it to understanding the moment-tomoment experiences of subjects responding to imaginaNve suggesNons that were aimed at temporarily altering perceived reality through deployment of phenomenological control. Our
intenNon was to generate novel hypotheses that could be tested in a future study. We presented three suggesNons individually to seven parNcipants, and then interviewed them about their experiences, with the qualitaNve findings generaNng four independent hypotheses. We repeated the process with six new parNcipants and found that the data supported three of the hypotheses. These were that while moderate responses involved goal-directed fantasies, high responses appeared to not; that responses scored low did not involve reports of acNve sabotage; and high and moderate responders do not appear to be aware that they are generaNng the suggested behavior and
cogniNons. These hypotheses will need confirming for generalizability with a quanNtaNve study. If confirmed, these results will bear on the mechanisms underpinning phenomenological control.</p
Investigating the potential for insect place learning using multi-scale CNNs
No full textMany insects, such as ants, are highly adept navigators, using learned visual information for route navigation and homing. They do this despite low resolution vision and small brains. Whether insects are also capable of more complex spatial cognition, such as pose independent place recognition, is an open question. In this study we first explored whether Convolutional Neural Networks (CNNs) of varying size are capable of a real world ‘ant’s eye’ place recognition task. We collected panoramic images from a set of 11 distinct places with variations in pose, time of day, weather and season. The CNNs were trained to categorise the places for input images of varying resolution. Whilst VGG16 performed best for all image resolutions, there was a general trend relating model size and image resolution to performance. Of the custom models, smaller models learn lower resolutions better than higher resolutions, and visa versa. We also found that resolutions 113 x 36 and 57 x 18 elicit the first or second best performance of the custom models, suggesting optimal performance for low computational processing lies between these two highlighted resolutions.</p
High-risk molecular features may eclipse genomic complexity in predicting chronic lymphocytic leukemia outcomes; UK clinical trial insights.
No full textHigh genomic complexity (HGC) is linked to poor prognosis in CLL, but its independent prognostic value remains uncertain amid emerging biomarkers. We analysed copy number alterations (CNA) in 495 untreated patients from (immuno)chemotherapy trials (CLL4, ADMIRE, ARCTIC), incorporating IGHV status, telomere length (TL), targeted sequencing and DNA-methylation subtypes. Patients harboured low (LGC, ≤2 CNAs; n = 334), intermediate (IGC, 3-4 CNAs; n = 97), or high (HGC, ≥5 CNAs; n = 64) genomic complexity. HGC associated with U-CLL (81%, p </p
[Book Review] Crude capitalism: oil, corporate power, and the making of the modern world market
No full textNo description supplied</p
Tolerance analysis and experimental validation of ROMI—a high-precision linear delta robot for microsurgery
No full textIn this paper we present the design of a tolerance analysis-based closed-loop system and a compensation framework applied to high-precision linear Delta robots. It considers the modelling of static and dynamic errors propagation arising from the structural tolerances and the end-effector’s positioning. This approach is combined with a closed-loop control system implemented using high-resolution optical encoders. The model is applied to the ROMI robot, a high-precision experimental Delta robot designed for microsurgical applications. Our simulation results reveal a theoretical home position error (the centre of the robot’s platform) of 1.9 mm, which is effectively compensated through kinematic calibration and a tolerance analysis-based closed-loop system. The proposed framework is evaluated experimentally through proof-of-concept experiments mimicking a microsurgical resection task conducted on a human peripheral nerve sample. The results from executing micrometre scale parallelogram and circular trajectories showed error reduction rates of 92.3% and 51.2% respectively, after five trajectory iterations. These findings confirm that manufacturing-induced errors can be consistently compensated using the proposed methodology, thus eliminating the need for ultra-high-precision machined components. This work establishes a practical and scalable pathway for designing more affordable high-precision robotic systems suitable for microsurgical and other high-precision applications.</p