21684 research outputs found
Sort by
Hearing the unspoken: thin slices theory and multiple instance learning for speech-based depression detection
No full textAbstract not currently available
Remote focusing to follow action potential transmurally in acute rabbit cardiac slices
Get PDFElectrical signals or action potentials (APs) originate in the pacemaker cells of the heart and travel through the organ in an orchestrated fashion. This electrical conduction governs mechanical contraction, making the heart an efficient pump. During amyocardial infarction (MI), oxygen deprived cardiac cells are rendered electrically inert as the scar is formed. This impairs the natural conduction pathways and can lead to life-threatening arrythmias. Therefore, understanding AP propagation in relation to the complex 3-dimensional tissue morphology of healthy and infarcted animal models is a pivotal step to establish diagnostic and therapeutic tools.
To indirectly investigate cardiac electrophysiology in large-scale intact tissue, optical mapping with voltage sensitive dyes (VSDs) is commonly employed [1]. Nevertheless, AP propagation in distinct cellular layers in depth below the epicardium is inaccessible with optical mapping. Conversely, two-photon fluorescence microscopy (2PM) allows cellular resolution and large tissue penetration depth with inherent optical sectioning [2]. In the ventricular wall of the heart, two directions of conduction can be considered: longitudinal, along the long axis of the cell and transmural, from the endocardium to the epicardium. Along the long axis of the cardiomyocytes, with 2PM, action potentials have previously been resolved as deep as 500 µm in distinct cellular layers of Langendorff-perfused rabbit hearts [3]. However, conventional microscopes cannot facilitate axial scanning fast enough to resolve action potentials and thereforetransmural cardiac conduction has not yet been investigated with 2PM.
In remote refocusing (RF), a remote objective and a lightweight mirror in its focal plane are introduced in the optical system [4, 5]. Consequently, the 3-dimensional optical copy of the sample can be probed vertically by rapid remote mirror actuation. RF is compatible with high numerical aperture 2PM systems and offers the temporal resolution necessary to resolve transmural action potential propagation over a large refocusing range [4, 5]. Studies employing RF in multiphoton microscopes focus predominantly on deep tissue neural function [6, 7]. Nevertheless, the brain is significantly less scattering compared to cardiac preparations. Sarcomere length of cardiomyocytes was measured utilising RF, but only cardiac structure, not electrical function was investigated [8].
Therefore, the aim of this work is, firstly, to develop a 2-photon microscope system enhanced with RF for rapid axial scanning and capable of probing action potential propagation along both longitudinal and transmural directions within deep myocardium. Our implementation of a remote refocusing module retrofitted to a commercial Scientifica 2-photon microscope achieves 250 Hz axial scanning over a range of 100 µm while maintaining under 5 µm axial resolution. The necessary system power efficiency and dispersion optimisation [9] is discussed: 22 mW average power with 15.9% throughput and a close to transform-limited pulse duration of 156 fs at sample is allowed by the system. Secondly, the full experimental pipeline including the preparation, viability evaluation and imaging of the cardiac preparation is established. Acute rabbit ventricular slice model adapted from [10] and prepared from hearts that have previously been used for other experiments (in line with the University’s strategy for reduction of animal research [11]) was tailored for 2P-RF imaging at room temperature. Live-dead staining (triphenyl tetrazolium chloride) of N=6 slices revealed no hypoxic core and gave a qualitative verification of slice viability in the central region of the preparations; slice contractility and action potential properties were characterised with optical mapping (CellOPTIQ system). The slices exhibit single-peak contractile traces throughout the entire preparation when stimulated at 0.3 Hz. Furthermore, in N=6 slices, the action potential duration was approximately 2 times shorter after increasing the concentration (from 15 mM to 30 mM) of 2,3-butanedione monoxime (BDM), an electromechanical uncoupler necessary for 2-photon imaging. The variation of AP duration from slice to slice at 30 mM BDM was characterised. Finally, the preliminary data to validate the use of our 2P-remote focusing to investigate cardiac AP propagation transmurally is presented. The time to acquire z-y planes in myocardium is reduced from minutes (conventional z-stack acquisition) to seconds with 122 Hz RF. When the remote refocusing module is bypassed, we resolve AP traces in electrically stimulated slices with rapid longitudinal galvanometric mirror scanning with SNR > 7 over the range of 120 µm in depth. Importantly, with the RF unit in place, action potential peaks were visible over approximately 60 µm range with static remote refocusing with SNR > 4. Therefore, we believe that the 2P-RF system presented will allow to resolve transmural APs with rapid axial scanning and enable a quantitative investigation how scar tissue impacts cardiac conduction in post-MI preparations
Essays on belief-driven macroeconomic volatility and expectation formation
Get PDFThis thesis investigates macroeconomic volatility through the lens of expectation formation, emphasising the role of belief-driven mechanisms. In particular, I explore the implications of incorporating Diagnostic Expectations (DE), a recent deviation from the standard rationality assumption, into macroeconomic models.
The first chapter embeds DE into a Small Open Economy framework à la Justiniano and Preston (2010), a benchmark model for analysing exchange rate dynamics. Recent studies show that DE generate excess volatility, short-term extrapolative behaviour and predictable shifts in investor sentiment, characteristics that align with puzzles in international macroeconomics, particularly excess exchange rate volatility and exchange rate disconnect (Obstfeld & Rogoff, 2000). For this reason, DE emerge naturally as a possible behavioural explanation for these phenomena. In this chapter, I leverage the international finance nature of the economy to study the interaction between DE and the exchange rate transmission channel, which is otherwise absent in a closed economy. I parameterise the model following the open economy literature and show that when the model is populated with diagnostic agents, the economy exhibits greater volatility vis `a vis the rational model. Moreover, DE introduce an amplification mechanism through shock extrapolation, which helps to qualitatively account for the excess volatility of the real exchange rate and its disconnection from fundamentals. The degree of departure from Rational Expectations (RE), captured by the diagnostic parameter, plays a central role in this extrapolation mechanism, with larger values amplifying the effect. I also use the model to assess the sensitivity of the results to different parameter values. The main finding highlights that economic openness and DE do not operate in isolation; rather, they amplify each other’s effects. In addition, I show that persistence mechanisms, especially interest rate smoothing, are essential for translating and intensifying the amplification effect of DE into short-run macroeconomic dynamics.
The second chapter expands the study of DE within macroeconomic models, now concentrating on the housing sector. Empirical evidence from the U.S. reveals that the housing market exhibits an unusually high degree of volatility, with survey-based expectations displaying biases that challenge the rationality assumption. In addition, tradition models often depend on volatile housing preference shocks to account for these fluctuations. In this chapter, I argue that the expectations channel plays a key role in driving housing market volatility. I incorporate DE into a Two-Agent New Keynesian (TANK) model featuring a housing and a banking sector to analyse the impact of this departure from rationality. I calibrate some parameters to the U.S. economy for the post-Volker - pre-Covid 19 pandemic period and estimate the remaining parameters using Sequential Monte Carlo methods. I find that DE reduce the volatility of the housing preference shock by more than one-third relative to RE, while still reproducing the observed housing market fluctuations. This result holds regardless of whether agents’ imperfect memory is based on recent or on three-year past experiences. When the expectations channel is removed, that is, when agents become rational, the model fails to generate the high volatility in house prices found in the data. These findings emphasise the importance of the expectations formation process for explaining a substantial part of the “unmodeled disturbances that can affect the housing market”, which Iacoviello and Neri (2010) attribute to a housing preference shock, and in shaping policy responses.
The third chapter extends the previous analyses by further demonstrating the effects of incorporating DE into macroeconomic models. Survey evidence, first presented by Coibion and Gorodnichenko (2015), sparked a broader discussion on deviations from the Full Information Rational Expectations (FIRE) framework (Fuhrer, 2018; Angeletos, Huo, & Sastry, 2021; Kohlhas & Walther, 2021). Specifically, Coibion and Gorodnichenko (2015) find that Forecast Errors (FE) and Forecast Revisions (FR) are predictable, suggesting that agents do not fully incorporate available information, a challenge to the FIRE hypothesis. In this chapter, I explore the impact of DE on the state-space structure of linear macroeconomic models and the resulting FE and FR across different horizons. In a three-equation specification, I derive testable expressions in terms of the model parameters and also generalise it to the case of larger models. I find that DE introduce predictability in the form of moving average (MA) processes. To assess whether expectation formation differs across agents, I analyse survey data from the Philadelphia Fed’s Survey of Professional Forecasters alongside policymakers’ forecasts from the Greenbook/Tealbook. The empirical results indicate that one-period-ahead FE generally follow the MA structures implied by DE, though evidence of overreaction appears only for GDP growth forecasts and primarily when including the post-pandemic period. Mixed results are, however, observed in the case of FR. For longer forecast horizons, FE include autoregressive components deviating from DE, whereas FR align more closely with DE-driven expectations, suggesting stronger revisions in the direction of the shock realisation. While these results provide insights into belief formation, they remain far from conclusive
Non-invasive AI-driven human activity recognition
Get PDFThe rapid proliferation of Internet of Things technologies, coupled with artificial intelligence driven applications, has revolutionised human activity recognition, enabling pervasive real-time monitoring across smart homes, healthcare, security, and ambient-assisted living environments. This transformation holds particular significance for healthcare systems, as radar-based recognition of physical and physiological activities facilitates continuous remote monitoring through invasive and non-invasive technologies, supporting personalised care and early intervention at scale. Traditionally, activity recognition systems have relied primarily on invasive or contact based devices, such as wearables and biosensors, which often lead to user discomfort, require frequent maintenance or charging, and risk non-compliance, especially among elderly individuals. Conversely, cameras, Wi-Fi, and radar are all treated as non-invasive sensing modalities; however, cameras raise serious privacy concerns and are constrained by lighting conditions, whereas Wi-Fi-based sensing suffers from multipath interference and spectrum-sharing challenges. Radar sensing emerges as a promising tool and privacy-preserving alternative with robustness to environmental variations. Despite these advantages, systems built on radar for activity recognition face significant challenges in real-world applications. This thesis addresses three critical challenges in radar-based human activity recognition: enabling non-intrusive recognition of both macro-level physical activities (e.g., falls, gait) and micro-level physiological signals (e.g., heart rate, respiration rate); data diversity and radar domain adaptation; and ensuring energy-efficient, privacy-aware edge deployment. The first contribution addresses the challenges of non-intrusive recognition of macro-level human activities and radar domain adaptation by developing a radar signal processing framework that transforms complex signals into four two-dimensional domain representations for robust activity recognition. By integrating domain-specific preprocessing with transfer learning, the framework improves adaptability across environments and reduces the complexity of raw signal data. Experimental results show up to 29.36% improvement in recognition accuracy compared to a baseline convolutional neural network, with transfer learning models achieving 96.03% on the primary dataset and demonstrating strong generalisation across two additional radar datasets. Building upon this, the second contribution focuses on finer-grained sensing, addressing the challenge of non-intrusive monitoring of micro-level physiological signals by extending the system to support radar-based extraction of vital signs such as heart rate and respiration rate. This is achieved using two radar modalities: ultra-wideband and millimetre-wave frequency-modulated continuous wave radar. A comprehensive analysis was conducted to evaluate the impact of varying distances and radar positioning configurations on the accuracy of vital sign extraction. The third contribution addresses the challenges of domain adaptation and energy efficiency by optimising transfer learning models for lightweight, energy-efficient, and privacy-aware deployment on edge devices. Using post-training quantisation and selective domain-model pairing, the system significantly reduces computational costs while maintaining high recognition performance across radar domains. Results indicate energy consumption as low as 0.42 mWh and response times of 1.32 seconds for 5-second activities, confirming its suitability for real-time, on-device monitoring. Additionally, the framework incorporates differential privacy techniques to strengthen local inference privacy with minimal loss in accuracy. Collectively, these contributions enhance the scalability, robustness, and efficiency of activity recognition systems, paving the way for non-invasive, AI-driven applications in healthcare and real-world environments
Motherhood in the shadows: shedding light on women's experiences of chronic invisible illness and maternity
No full textAbstract not currently available
Sound before picture: towards a sound led videographic criticism
Get PDFVideographic criticism has gained a significant foothold within the academic film studies community as a form of research and publication. The number of journals that now accept video essays as scholarly publications has grown over the last decade, and videographic works are now finding exhibition opportunities at film festivals and conferences, often bridging or balancing on the point between artistic and academic practice. Video essay courses and modules are also now part of undergraduate and postgraduate teaching at universities in the UK, Europe, and the United States. A growing body of critical writing and discussion has developed around the scholarly video essay, though there is a noticeable absence of discussion focused on the technical craft and skills involved in the making of videographic work, especially in relation to technical sound skills.
This thesis seeks to understand the role of aesthetic and technical sound skills in videographic criticism. Through analysis of published works, critical discussion, and practical engagement in the making of video essays, it will consider the contribution technical sound skills might make to this developing form of academic engagement. It will also seek to identify a means by which critical discussions and critical engagement with the technical aspects of videographic making, and specifically technical sound literacy, might be developed within the wider community of videographic critics
Empirical essays in the economics of education
Get PDFThis thesis contributes to understanding the education of disadvantaged children within the field of Economics, with a focus on rural China, where educational inequalities are prominent. Chapter 1 examines the relationship between parental migration and children’s cognitive abilities and school engagement. The findings show that while parental migration does not significantly impact cognitive abilities, it negatively affects school satisfaction for both boys and girls, with a more pronounced effect on boys. Boys’ school engagement also suffers due to parental absence. Chapter 2 presents a survey report. The survey focuses on ethnic minority children who were potentially exposed to the "One Village, One Preschool" (OVOP) program. The survey gathered data on family background, parental care, school engagement, personality, and peer relationships. The results indicate that participation in the OVOP program enhances social skills, self-control, and learning habits, which helps explain the academic performance gap between participants and non-participants. Chapter 3 evaluates the impact of the OVOP program on ethnic minority children’s academic performance and socioemotional skills using academic records and survey data. The findings suggest that the OVOP program significantly improves academic outcomes and enhances socio-emotional skills, including task performance, emotional regulation, social engagement, and open-mindedness
Development participation and adoption intention of ICT for informed decision-making in urban public services: dashboard for Jakarta traffic police patrolling allocations
Get PDFThis thesis investigates the development participation and adoption intention of Information and Communication Technology (ICT) for informed decision-making in urban public services, using a dashboard as an ICT example and Jakarta Traffic Police’s patrolling allocation as the use case of decision-making. This research addresses the gaps in understanding Data-Driven Decision-Making (DDDM) in the public sector and the limited study of government-to-employee (G2E) ICT adoption. It also aims to investigate the underexplored role of user participation in the development and adoption of government ICT. Furthermore, this research underscores the need for qualitative insights to inform the development of quantitative hypotheses, particularly where traditional technology acceptance models like TAM and UTAUT may overlook key public sector contextual factors.
The study is divided into two phases: requirement elicitation and adoption intention examination. In the first phase, semi-structured interviews with Jakarta traffic police were conducted to explore their decision-making processes within a Data-Driven Decision-Making (DDDM) framework and identify preferred features for a user-centred dashboard, leading to the development of a tailored prototype. In the second phase, semi-structured interviews were conducted to develop original hypotheses, focusing on public sector-specific contextual factors. These were then quantitatively tested to examine the factors influencing the traffic police's behavioural intention to adopt the dashboard.
The novel contributions of this research include the development of the original contextual hypotheses and identifying new motivational factors influencing government employees’ ICT adoption, such as interoperability expectations, social incentives, and user experience-related factors. The research also highlights the importance of user participation in enhancing the intention to adopt ICT tools in the public sector. The study provides practical insights for improving public service delivery through data-driven tools and contributes to refining ICT adoption theories in public sector contexts while offering evidence-based recommendations that integrate both technological advancements and human insights to foster more resilient and adaptive decision-making
Reconfigurable arbitrary basis mode sorter and structured light processor based on an integrated photonic platform
No full textAbstract not currently available
Cultural constraints in digital adaptation: state ownership, agents, and newspaper organisations in China
Get PDFThis study examines the digital adaptation of China’s state-owned newspaper firms, with a particular focus on the cultural constraints shaping their organisational capabilities for digital growth. Drawing on theories from organisational studies, media management, and media innovation, it investigates ownership as a key structural factor influencing cultural dynamics, which in turn affects the innovation capabilities of Chinese newspaper firms.
Through a multi-case study of three Beijing-based national industry newspapers, this research identifies three common cultural phenomena: self-identification as state media, symbolic compliance in policy engagement, and the accepted norm of “co-creation” between journalists and advertisers. Despite their shared state ownership, each newspaper firm exhibits distinct organisational cultures shaped by agent owners—supervisory entities responsible for exercising ownership rights on behalf of the state. By embedding their own institutional priorities and interests into organisational practices, these agent owners refract the influence of state ownership, generating varied cultural dynamics across Chinese newspaper firms. Building on this, the study further explores the motivations behind Chinese newspaper firms’ innovative behaviours, identifying three primary drivers—self-drive, state compliance, and change aversion—that either facilitate or hinder digital adaptation. The interplay of these motives, with state ownership playing a significant role, reveals a misalignment of values, goals, and expectations among policymakers, agent owners, press management, and practitioners. These cultural constraints have been limiting the innovation capabilities of Chinese newspaper firms in their pursuit of digital transformation and genuine engagement with the state-led media convergence strategy.
This study challenges the conventional view of China’s state-owned media system by critically uncovering the underexplored yet pivotal role of agent owners. It further advances media innovation research by proposing a novel motivation-based framework to decode organisational cultures and behaviours within Chinese newspaper firms, drawing on valuable empirical data. The findings call for further academic and regulatory inquiry into the power boundaries of agent owners and their conflicts of interest with Chinese newspaper firms, offering insights to mitigate agency loss in China’s media governance and address the institutional constraints on media innovation
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! 👇