42337 research outputs found
Sort by
Scalable random forest for plug-and-play myoelectric control
Get PDFMyoelectric control systems translate electromyography (EMG), the non-invasive muscle
electrical signals detected through cutaneous electrodes, into control commands for external
devices, enabling human-machine interactions across diverse applications. Despite remarkable
advancements, myoelectric control systems still face challenges, primarily due to the
inherent EMG variability stemming from multiple confounding factors, including inter-user
physiological differences, long-term signal drift, dynamic arm positions, and noise contamination.
While state-of-the-art decoders have shown excellent performance in controlled settings,
their low robustness, high computational complexity, lack of explainability, and heavy reliance
on extensive calibration data limit their practical implementation in real-world applications.
The doctoral research presented in this thesis aims to find a practical decoder for myoelectric
control that addresses the above-mentioned challenges. Specifically, this thesis presents a
systematic investigation of simple but powerful random forest (RF) as an alternative decoder,
unlocking the underutilized strengths of RF models to develop robust, adaptable, inter-user
generalizable, explainable, easily parallelizable, and computationally efficient myoelectric
control systems. We introduce five key innovations: (1) explainable deep forest models for
high-density EMG (HD-EMG) processing, combining inherent noise resilience with physiologically
consistent model interpretations. (2) a pre-trainable RF framework enabling one-shot
personalization by grafting, pruning and appending decision trees; (3) a self-calibrating RF
framework that progressively adapts to long-term (up to 5 weeks) EMG variations without
labeled calibration data; (4) posture-invariant RF models demonstrating inherent robustness
to arm position changes; and (5) scalable RF that achieves ≈500× model size compression
with negligible accuracy loss. Extensive experiments including intra-day/long-term testing
scenarios, with data from 106 participants, were conducted to validate above innovations.
This research establishes scalable RF as a viable and practical foundation for plug-andplay
myoelectric interfaces, simultaneously satisfying four key requirements often considered
mutually exclusive in the field: (1) high classification accuracy, (2) computational efficiency,
(3) physiological explainability, and (4) minimal calibration demands. The proposed solutions
bridge the critical gaps between laboratory demonstrations and real-world settings
A systematic review of therapist reflective functioning: characteristics, measurement, construct validity and outcome; & Exploring supervisor’s experiences when delivering psychological supervision: an interpretative phenomenological analysis; & Exploring supervisor’s experiences when delivering psychological supervision: an interpretative phenomenological analysis
No full textBACKGROUND:
This thesis portfolio comprises two main sections. The first involves a systematic review exploring studies that measured therapists' mentalisation capacity, also commonly referred to as reflective functioning (RF). As research into the importance of therapists' RF is relatively new, this review remained exploratory, looking at the characteristics of included studies, RF measurement tools used, the construct validity of such tools, and whether therapist RF appears to influence patient outcomes following psychotherapy. This may help us understand whether therapist RF is an important common factor and indeed change mechanism across psychotherapy models. The second section is an empirical study following on the theme of mentalisation theory but from a supervisory perspective. Supervising is an emotional and relational experience, often with a strong restorative and regulatory focus. Research looking at the emotional and interpersonal demands placed on supervisors is lacking. It is possible that supervisory interactions could also be underpinned by mentalisation processes. This could be from a function of supervision perspective where supervisors may attempt to help supervisees sustain mentalising about their work but also the process of the supervisors’ own mentalising - recognising the emotional work that supervising entails. This research, therefore, sought to explore the emotional and interpersonal experiences of supervisors delivering supervision from a mentalisation theory perspective. This may help us to understand the importance of mentalisation processes within this forum and help inform clinical practice and supervision training models.
METHODS:
The review involved a systematic search of three online databases and reference lists to identify relevant articles, determined by strict eligibility criteria. Fourteen articles were identified, which were synthesised and assessed for risk of bias by the standard quality assessment criterion for evaluating primary research papers from a variety of fields (Kmet, 2004). For the empirical project, eight registered psychologists completed semi-structured interviews relating to their emotional and interpersonal experiences of delivering supervision. Interviews were recorded, transcribed verbatim, and analysed using Interpretative Phenomenological Analysis (IPA).
RESULTS:
The systematic review revealed large heterogeneity across studies in terms of research design, RF measurement tools used, and treatment modality, meaning cross-study comparison was difficult. There is preliminary evidence of construct validity across RF measures. Therapist RF appears relevant as a potential change mechanism within psychotherapy processes. IPA from the empirical project revealed four main themes: ‘supervisor vulnerabilities’, ‘ the importance of reciprocity’, ‘supervision is relational’, and ‘a mentalising stance’.
DISCUSSION:
Studies included in the review highlight the role therapist RF could have as a common factor and as an important change mechanism regardless of psychotherapeutic orientation. This has direct implications in the application of therapy models but also in the selection and training of mental health practitioners. Key themes such as supervisor vulnerabilities give rise to supervisors’ seeking
validation of skills and opportunities to demonstrate competence. This is discussed in relation to self-validation literature. Learning from the supervisee appeared to ensure that the supervisor felt useful and validated, valuing the importance of reciprocal learning within the supervisory relationship. The findings also highlight that mentalisation underpins almost all interpersonal interactions, especially in a relational space like supervisory relationships. Evidence of supervisors’ adopting a mentalising stance is explored, alongside a critical appraisal of the research process
A complete framework for agile quadruped locomotion: integrating real-time control, planning, and perception in multi-contact environments
No full textReal-time synthesis of legged locomotion manoeuvres in challenging environments, such as industrial staircases with distinct contact surfaces, remains an unresolved problem. These environments, known as multi-contact environments, require the simultaneous determination of footstep locations several steps ahead while generating whole-body motions close to the robot's operational limits. The practical constraint of state estimation and perception errors necessitates rapid re-planning of motions. With traditional model-based methods, this constraint prevents using a single large optimisation to solve the entire locomotion problem, leading to problem decomposition into smaller, more manageable sub-problems to meet computing time requirements.
However, such decomposition introduces issues such as loose coupling between sub-problems due to varying models, constraints, and horizons, highlighting the need for an efficient architecture. Decomposition, particularly at the contact level to manage complexity, exacerbates the challenge of addressing combinatorial problems in multi-contact environments.
We propose a fully decomposed control architecture to address the locomotion problem, leveraging mixed-integer optimisation with conservative assumptions as a planner to select only the contact surfaces while continuously updating footstep positions and leg trajectories within these surfaces. The approach was initially validated on the quadruped robot Solo and later extended to include real-time perception with the larger industrial robot ANYmal-B. Additionally, we explore integrating learning-based methods along with trajectory optimisation frameworks to leverage the strengths of both approaches and enhance the planner's performance
Designing a motivating home exercise app for children
No full textPaediatric physiotherapy deploys a range of methods to rehabilitate and recover movement for children who have been injured, had surgery, or have a disability that affects their motor function. This includes home exercise programs (HEPs) which are a curated set of exercises to be completed at home as part of physiotherapy treatment. Although considered a pillar of treatment within the field, adherence to HEPs has been reported as low as 50%, with patients and their families reporting barriers such as forgetfulness, lack of time, and lack of enjoyment. The use of technology to deliver physiotherapy treatment greatly increased during the COVID-19 pandemic, resulting in a better understanding of how mobile technologies can be used as a tool to reach patients at home. Recent work has started to use this experience to inform the design of mobile health (mHealth) interventions that target these barriers to adherence to treatments or medications within physiotherapy and adjacent fields. However, previous reviews have reported that there is a lack of evidence-based design with these solutions, and have mapped out safety concerns that result from a lack of consultation with experts and target users.
This thesis investigates the design, development, and evaluation of an app that delivers and tracks progress through a goal-based home exercise program (HEP) for children aged 7-11. Through the application of a participatory design method with a group of 24 children and 2 expert physiotherapists, informing functionality, behaviour change, and safety, this research provides requirements and designs for an app that encourages and motivates adherence. Workshops were used to design fundamental functionality within the app with children, including applying behaviour change theory through a rewards system, goal-setting functionality, and a character that motivates the user to exercise. Feedback from designers and testers showed that genuine participation from child designers during the creation of complex interventions is possible and that a design environment with a `least adult’ approach strengthened the adult-child relationship in these workshops.
The results of a 4-week evaluation with 11 additional healthy child athletes generated initial positive findings for the guided goal-setting functionality created through the observation of gameplay with child designers. The athletes could set and track progress towards goals and the 5 athletes from the user group that were interviewed reported that they were motivated to exercise. The rewards system, built on a foundation of fairness defined by child designers offers a promising design for future work, with positive feedback from athletes. Finally, the characters, designed by children with speech modelled after the child designers were particularly powerful in encouraging athletes to return to the app between sessions and also served as a tool for generating discussion or participation in the HEP with family and friends.
The primary contribution of this work is the design and requirements for an app that can deliver, and encourage adherence to, an HEP for children. Additionally, the child-centred design process in this work clearly showed benefits for both the child designers’ and the child evaluators’ enjoyment of the app. Reflecting on the results of the evaluation with physiotherapists, HomeExerciseBuddy has potential for prescription as part of physiotherapy practice to encourage adherence to HEPs. The initial results gathered from healthy athletes were reassuring for the usability of the application and the potential for a digital solution to this adherence problem with children was exciting to expert physiotherapists
Models of corticostriatal synaptic plasticity and plateau potentials in striatal projection neurons
No full textIn this thesis we studied synaptic plasticity and neuronal computation in single striatal projection neurons (SPNs), which have a major role in goal-directed learning. Goal-directed or reward learning means to learn, based on sensory information from the body and the environment, to select actions out of all the behavioral repertoire that lead to obtaining a goal or reward (such as food or water). In mammals, all the behavioral motor repertoire is under constant, tonic inhibition, and the direct-pathway SPNs (dSPNs) select (disinhibit) goal-obtaining actions. The learning process is guided by the neuromodulator dopamine which signals the positive or negative outcome of an action. The synapses from cortical neurons onto the dSPNs, called corticostriatal synapses, are responsive to dopamine signals, and can strengthen and weaken based on the (positive or negative) action outcome. This promotes or discourages future actions in the same or similar sensory context.
Within a collaborative computational modeling effort, we studied the biochemical circuitry in the corticostriatal synapses with multiscale modeling and simulations. This circuitry in the corticostriatal synapses responds to neuromodulatory signals and controls the expression of synaptic plasticity. Multiscale modeling and simulations enable studying a system at multiple temporal and spatial scales, and integrating the results across the different scales. Based on molecular dynamics simulations of the enzyme which transduces extracellular neuromodulatory signals into an intracellular second messenger molecule, and Brownian dynamics simulations of regulator molecules binding to the enzyme, we constructed a kinetic model of the enzyme-based signal transduction network. The kinetic model showed that two co-occuring neuromodulatory signals, a dopamine peak and an acetylcholine pause, are required to produce the second messenger and thus enable strengthening of corticostriatal synapses onto dSPNs, and that only the dopamine signal is not enough.
Next, we developed a local, calcium- and reward-dependent learning rule based on what is known about the biochemical circuitry of corticostriatal synapses onto dSPNs. We show that with this biologically-based learning rule, single SPNs can learn to solve the nonlinear feature binding problem (NFBP), a computationally hard problem representing the class of linearly nonseparable tasks. This result suggests that different, unrelated or partially related stimuli that require executing the same action to obtain a goal, can use the same SPNs responsible for selecting that action, and that a single SPN can reliably distinguish between similar stimuli.
The solution of the NFBP with the aforementioned learning rule relies on supralinear dendritic voltage elevations called plateau potentials. Experimentally, plateau potentials are all-or-none events, a property crucial for performing nonlinear computations required to solve the NFBP. However, computational models of plateau potentials often produce graded voltage elevations. We analyzed and compared existing plateau potential models, and found that long-lasting glutamate spillover in the extrasynaptic space robustly produces all-or-none plateau potentials by activating extrasynaptic N-methyl-D-aspartate (NMDA) glutamate receptors. This suggests that glutamate spillover may be a mechanism for generating all-or-none plateau potentials in vivo, as well.
In summary, the findings presented in this thesis advance our understanding of the role of single dSPNs in goal-directed learning, the biophysical mechanisms involved in performing their nonlinear computations, and the neuromodulatory signals necessary to produce synaptic strengthening and thus implement goal-directed learning
The equitable adaptation of the law to climate change: the case of informal settlements
Get PDFInformal settlements are one of the most vulnerable communities to the effects of
climate change, requiring climate action to increase the adaptive capacity and
resilience of their inhabitants to cope with the impacts of climate change. These
measures involve legal and institutional adaptation processes; thus, this thesis aims
to justify the ‘equitable adaptation of the law’ as a helpful method to improve the
protection of informal settlers from climate change impacts. In particular, the analysis
focuses on tackling the obstacles that prevent informal dwellers from coping with
climate change effects derived from property and urban law by using the equitable
adaptation of the law. Finally, aiming to have more concrete conclusions on the utility
of the equitable adaptation of the law, this thesis applies the framework developed in
the fields of property and urban law for equitably adapting the law to the cases of Rio
de Janeiro and Delhi
Visual connection: the role of video-sharing sites in shaping minority stress experiences for LGB individuals in Hong Kong and Macau
Get PDFINTRODUCTION:
Sexual minority communities, particularly those identifying as lesbian, gay, and bisexual (LGB), encounter heightened minority stress (MS) that negatively impacts their mental health. In Chinese communities, these stressors are uniquely intensified by cultural norms, traditional values, and social expectations, leading to increased social stigma and familial pressure not as prevalent in Western contexts. Despite the potential of video-sharing sites (VSS) to alleviate MS, the specific challenges faced by LGB individuals in these environments necessitate targeted research to understand and mitigate their unique experiences effectively.
BACKGROUND:
Existing research establishes the buffering effect of social identity (SI) against stress and links a negative attributional style (NAS) with increased psychological distress. The nuanced impact of use of LGBT video-sharing sites (ULV) on MS, primarily through the interplay of SI and NAS, remains to be fully elucidated. This gap highlights the need for a focused investigation into the potential mediating roles of SI and NAS in the relationship between ULV and MS reduction.
AIM AND MODEL DEVELOPMENT:
This study aims to bridge this gap by investigating the mediating roles of SI and NAS in how ULV influences MS among LGB individuals. A robust literature review has developed a conceptual model that proposes that ULV influences MS through a serial mediation of SI and NAS. This model hypothesises that ULV can enhance SI, which subsequently reduces NAS, thereby lowering MS.
METHODS:
The study employed a quantitative research design, gathering data from 220 Hong Kong and Macau LGB individuals. Participants were surveyed on their ULV, experiences of MS, levels of SI, and NAS. The survey included measures such as the Minority Stress Scale, Social Identity Scale for LGB individuals, and a Negative Attributional Style questionnaire. Multiple regression analyses and Hayes’s process for testing mediation hypotheses were key statistical tests used to explore the relationships between ULV, MS, SI, and NAS.
FINDINGS:
Results indicated a significant negative correlation between ULV and MS, demonstrating that higher ULV is associated with lower levels of MS. The analysis suggested that SI and NAS may serve as mediators in this relationship, with SI identified as a crucial mediator. Specifically, the findings indicate that ULV is positively associated with SI, which in turn is negatively associated with NAS, potentially leading to a reduction in MS. The mediating effect of NAS was significant only when preceded by an enhancement in SI, highlighting the importance of strengthening SI for reducing MS.
DISCUSSION:
This study expands our understanding of the mechanisms through which LGBT VSS may support LGB individuals in mitigating MS. It emphasises the critical role of fostering SI and provides evidence for the serial mediation model developed from the literature review. The findings suggest that future research explore additional factors that may influence the ULV-MS relationship and consider the integration of LGBT VSS into mental health support strategies for the LGB community.
CONCLUSION:
By highlighting the potential of LGBT VSS in associating with reduced MS through the enhancement of SI and the subsequent reduction in NAS, this research underscores the importance of such platforms in supporting the mental health of LGB individuals. It calls for the adoption of inclusive mental health policies and the inclusion of VSS in outreach and support strategies, offering a comprehensive framework for future research and practice aimed at improving the wellbeing of sexual minorities
Enhancing corporate bankruptcy and financial distress prediction: a multi-method approach with profiling and network analysis
No full textThis thesis encompasses three interrelated studies on the prediction of corporate bankruptcy and financial distress, a critical area for various stakeholders including businesses, financial institutions, investors, regulatory bodies, auditors, and academics.
The first study presents a comprehensive survey, classification, and critical analysis of the literature on corporate bankruptcy and financial distress prediction. It covers definitions, prediction methodologies, data pre-processing, feature selection, model implementation, performance criteria, and evaluation methodologies, providing a critical analysis to inspire future research directions.
The second study introduces a novel approach to bankruptcy prediction by leveraging company relational information through complex network analysis. Using board of directors’ networks, node embeddings and communities, the study devises new corporate governance drivers to enhance prediction accuracy. Empirical results from UK companies listed on the London Stock Exchange demonstrate that these network-based drivers significantly improve prediction performance.
The third study proposes a two-step methodology for bankruptcy prediction utilising companies’ financial profiles. Initially, unsupervised cluster analysis generates group financial profiles from accounting information. Subsequently, supervised artificial intelligence models predict bankruptcy based on these profiles. This approach, tested on data from UK companies listed on the London Stock Exchange, shows improved prediction performance and enhanced model interpretability.
Together, these studies contribute to the advancement of bankruptcy prediction methodologies by integrating comprehensive literature analysis, innovative network-based drivers, and a two-step financial profile-based prediction approach
Potential for interactive EPCs for Scotland
Get PDFThis project considers whether it would be beneficial to incorporate data or functionality into Energy Performance Certificates (EPC) in Scotland. Three broad levels of interactivity are proposed that could allow householders to better assess potential retrofit measures and may prompt households to undertake energy efficiency measures or switch to clean heat systems
Fluorescence lifetime imaging with distance and ranging for biomedical endoscopic applications
Get PDFEndoscopic cameras play a vital role in medical diagnostics as they provide a minimally invasive way to image inside the body. Conventional cameras, however, are reliant on the use of white light which does not offer any additional diagnostic information over cues from colour and intensity.
Fluorescence lifetime imaging (FLIm) is a technique that can enhance endoscopy by providing differentiation between tissue types without the need
for biomarkers. This could be particularly valuable, for example, in the identification of cancerous tissue margins pre-intervention.
Time resolved arrays which incorporate single photon avalanche diodes (SPADs) are a prime candidate for FLIm as they offer very fast and precise
measurements of individual photon arrival times. In addition, they are able to obtain high speed images, even in low light conditions such as typically
observed in FLIm of endogenous fluorophores in the human body.
Another time resolved technique enabled by SPADs is time of flight (ToF) imaging which allows the distance to a target to be determined. ToF based
distance mapping could provide crucial depth perception which FLIm images inherently lack. In the case of endoscopic imaging, this allows for greater
camera and instrument control which in turn mitigates the risk of tissue damage during surgical procedures.
This research presents a combined fluorescence lifetime imaging with distance and ranging (FLImDAR) technique, realised through the use of SPAD
arrays. The FLImDAR modality has strong applications in clinical endoscopic imaging and thus is the focus of this research. Similar previous studies have
investigated depth measurements using deep tissue fluorescence which requires the addition of fluorescent dyes. This work, however, targets surface
auto-fluorescence.
Firstly, a computational model is used to examine and refine various algorithms which can be implemented to perform FLImDAR. This model is
used to decide the final FLImDAR technique, as well as investigate the impact of signal noise. Next, FLImDAR is demonstrated experimentally using
fluorescent polymer targets. Three SPAD sensors (MegaFrame, Quanticam and Endocam) are tested for compatibility with the FLImDAR modality, each
showcasing different strengths and limitations. To further prove FLImDAR’s ability to perform in low photon level auto-fluorescent applications, ovine and cancerous human pulmonary targets are imaged. The final part of this study regards performing FLImDAR with a miniaturised imaging system. The SPAD sensor, Endocam, is characterised prior to integration into a handheld camera. Using the handheld camera, FLIm videos of ovine tissue are captured as well as long exposure FLImDAR images.
With further advances in detectors, sensor architecture and photon budgets, this research indicates the significant performance improvements that will enable practical clinical application of the FLImDAR technique
Access Repository Dashboard
Do you manage Edinburgh Research Archive? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!