Date | Day | Time | Week no. | Presenter | Affil. | Dept. Contact | Title |
|---|---|---|---|---|---|---|---|
2024-09-13 | Friday | 2pm | 1 | Nastaran Sharifian | Galway University | Kevin Burke | Automating functional data analysis in real time: an application to pressure sensor data in the treatment of venous leg ulcers |
2024-09-20 | Friday | 2pm | 2 | Laurent Hébert-Dufresne | University of Vermont | James Gleeson | Contagion models that challenge the linear relationship between exposure and transmission |
2024-09-27 | Friday | 2pm | 3 | Graham Benham | University College Dublin | Doireann O'Kiely | Wave-driven propulsion |
2024-10-11 | Friday | 2pm | 5 | Vincent Labatut | UL | Padraig.MacCarron | Continuous Average Straightness in Spatial Graphs |
2024-10-18 | Friday | 2pm | 6 | Lennon Ó Náraigh | University College Dublin | Eugene Benilov | A mathematical model and mesh-free numerical method for contact-line motion in lubrication theory |
2024-10-31 | Thursday | 2pm | 8 | Tim Myers | Centre de Recerca Matemàtica, Barcelona | James Gleeson | Mathematical modelling of the adsorption of environmental contaminants |
2024-11-08 | Friday | 2pm | 9 | Jochen Einbeck | Durham Univerity | Shirin Moghaddam | Statistical modelling in radiation biodosimetry |
2024-11-15 | Friday | 2pm | 10 | Susana Gomes | University of Warwick | Doireann O'Kiely | Control of falling liquid films using a hierarchical model approach |
2024-11-29 | Friday | 2pm | 12 | Edouard Bonneville-Online | Leiden University Medical Centre | Shirin Moghaddam | Competing risks and multiple imputation of missing covariates |
2024-12-06 | Friday | 2pm | 13 | Daniel Andre | Cranfield University | Cliff Nolan | Multistatic Synthetic Aperture Radar Imaging and Coherence |
AY 2024-2025 Semester 1 Seminar Schedule
Timetable
Abstracts
As seminar abstracts become available from the speakers this page will be updated.
Seminar week 1 by Nastaran Sharifian
Date: 2024-09-13 at 2pm
Speaker: Nastaran Sharifian (Galway University)
Host: Kevin Burke
Title: Automating functional data analysis in real time: an application to pressure sensor data in the treatment of venous leg ulcers
Abstract: Venous Leg Ulcers (VLUs) afflict approximately 11.5 million individuals globally, stemming from impaired leg vein function. Compression therapy is the conventional treatment for VLUs, but ensuring appropriate pressure remains a challenge. The novel medical device, Tight Alright developed by FeelTect, can measure bandage pressure 20 times per second at three relevant physiological points on the leg, ensuring effective venous return along the length of the leg and promoting VLU healing. The Tight Alright connected health platform provides measurement of sub-bandage pressure when applying compression therapy, ensuring safe and efficacious target pressure is consistently delivered. These pressure data are transmitted wirelessly (via Bluetooth) to a user-friendly Mobile App, leading to multivariate real-time sensor data. Pressure data are available from the same individual using the pressure sensing device and app under two pressure settings while performing three exercises for 60 seconds each.
Leveraging Functional Data Analysis (FDA), we smoothed these data to derive a finite-dimensional representation for model fitting, facilitating Multivariate Functional Principal Component Analysis (MFPCA). MFPCA, akin to traditional Principal Component Analysis (PCA) but tailored for multivariate functional data, decomposes variability into orthogonal functional principal components (FPCs), enabling dimension reduction and visualisation. Each pressure wave is attributed a multivariate score for each component of variation, and these scores are then modelled using longitudinal data analysis to investigate pressure change over time and compare between exercise states.
We develop an algorithm to automate pressure wave detection, segmentation, registration, and MFPCA, culminating in multivariate principal component scores. These scores compute variation across the three sensor positions for each wave. We use longitudinal data analysis to model these scores over time to find differences between exercise and starting pressure.
This approach accommodates the influx of pressure sensor data, ensuring adaptability to new datasets and facilitating real-time analysis for individuals utilising the monitoring device so as to harness and harmonise these multimodal data into useful user facing visualisations on the app. This will ultimately advance monitoring and treatment decision making in VLU wound treatment, which will advance the use of connected-health technology with precision compression therapy to expedite wound healing.
Keywords: Venous Leg Ulcers, Compression Therapy, Pressure Sensor Data, Multivariate Functional Principal Component Analysis
Seminar week 2 by Laurent Hébert-Dufresne
Date: 2024-09-20 at 2pm
Speaker: Laurent Hébert-Dufresne (University of Vermont)
Host: James Gleeson
Title: Contagion models that challenge the linear relationship between exposure and transmission
Abstract: Models of contagions attempt to reduce complex global dynamics (like a pandemic) to a set of simple local interactions (pairwise transmissions). To do so, it is common to assume that the force of infection on a susceptible individual is linearly proportional to its exposure to the contagion. Here, I will tell two stories of models that implicitly challenge this assumption by acknowledging that the context of exposure matters. First, we will look at heterogeneous patterns of transmission occurring when infection risk varies by orders of magnitude in different settings, like indoor versus outdoor gatherings in the COVID-19 pandemic. Second, we will stop assuming that epidemics happen in a vacuum and consider interactions through co-infections, like COVID-19 and influenza. Spoiler alert: Both stories will show an induced superlinear force of infection during the emergence of a new outbreak when we are unaware of the full context of the exposure and transmission events. These results potentially have important consequences in a wide range of modelling scenarios as they illustrate that superlinear dynamics can emerge from unobserved covariates even when the dynamics is otherwise linear.
Seminar week 3 by Graham Benham
Date: 2024-09-27 at 2pm
Speaker: Graham Benham (University College Dublin)
Host: Doireann O’Kiely
Title: Wave-driven propulsion
Abstract: Wave-driven propulsion occurs when a floating body, driven into oscillations at the fluid interface, is propelled by the waves generated by its own motion. Wave-driven propulsion has been observed in the case of the waves generated by a honeybee trapped on the surface of water, in the case of “SurferBot”, a centimeter-scale interfacial robot that was inspired by the stricken honeybee, and at much larger scales, in the case of the waves generated by jumping up and down on a canoe, also known as “gunwale bobbing”.
In this seminar I will present a new theory for wave-driven propulsion based on coupling the equations of motion of a floating raft to a quasi-potential flow model of the fluid. Using this model, expressions are derived for the drift speed and propulsive thrust of the raft which in turn are shown to be consistent with global momentum conservation. The validity of the model is explored by describing the motion of SurferBot, demonstrating close agreement with the experimentally determined drift speed and oscillatory dynamics. The efficiency of wave-driven propulsion is then computed as a function of driving oscillation frequency and the forcing location, revealing optimal values for both of these parameters which await confirmation in experiments.
Seminar week 5 by Vincent Labatut
Date: 2024-10-11 at 2pm
Speaker: Vincent Labatut (UL)
Host: Padraig.MacCarron
Title: Continuous Average Straightness in Spatial Graphs
Abstract: The Straightness is a measure designed to characterize a pair of vertices in a spatial graph. In practice, it is often averaged over the whole graph, or a part of it. The standard approach consists in: 1) discretizing the graph edges; 2) processing the vertex-to-vertex Straightness considering the additional vertices resulting from this discretization; and 3) averaging the obtained values. However, this discrete approximation can be computationally expensive on large graphs, and its precision has not been clearly assessed. In this work, we adopt a continuous approach to average the Straightness over the edges of spatial graphs. This allows us to derive 5 distinct measures able to characterize precisely the accessibility of the whole graph, as well as individual vertices and edges. Our method is generic and could be applied to other measures designed for spatial graphs. We perform an experimental evaluation of our continuous average Straightness measures, and show how they behave differently from the traditional vertex-to-vertex ones. Moreover, we also study their discrete approximations, and show that our approach is globally less demanding in terms of both processing time and memory usage.
Seminar week 6 by Lennon Ó Náraigh
Date: 2024-10-18 at 2pm
Speaker: Lennon Ó Náraigh (University College Dublin)
Host: Eugene Benilov
Title: A mathematical model and mesh-free numerical method for contact-line motion in lubrication theory
Abstract: In gas-liquid flows, a contact line refers to the intersection between the gas-liquid interface and a boundary wall. If the fluid is in motion, then so is the contact line. This apparently contradicts the no-slip boundary condition of fluid dynamics, and gives rise to a singularity in the equations of motion. Such a paradox signals missing physics. A problem is that the missing physics is on the nanoscale, whereas the typical fluid dynamical applications are on the micron or millimeter scales. It becomes necessary in practice to parametrize the missing physics. In this talk, we give an overview of two classical parametrizations which are extremely useful for two-phase flow simulations. This part of the talk will be more of a review of the state-of-the-art. Then, we will introduce a novel regularization of the contact-line singularity problem, valid in the case of thin-film flows.
Seminar week 8 by Tim Myers
Date: 2024-10-31 at 2pm
Speaker: Tim Myers (Centre de Recerca Matemàtica, Barcelona)
Host: James Gleeson
Title: Mathematical modelling of the adsorption of environmental contaminants
Abstract: It is well-documented that it is now virtually impossible to reach internationally agreed-upon energy and climate goals without active removal of environmental contaminants, combined with substantial emission reductions across all sectors. Whilst the solution can only come through a number of technologies, one of the most common and versatile methods for the capture of contaminants from a fluid is via column adsorption. Adsorption columns are employed in the removal of greenhouse gases, volatile organic compounds, emerging contaminants and PFAs, also in water treatment, biogas cleansing and the purification of biopharmaceutical products. They are relatively easy to introduce into an industrial chain and may be applied to both liquids and gases. Consequently, they are a key tool for environmental remediation. In this talk I will discuss recent work of the Environmental Mathematics group in Barcelona. The EM group specialises in the development and analysis of mathematical models of physical processes related to environmental issues. Mathematical solutions are particularly important since they provide explicit relations for system parameters and so lead to an understanding of the physical process not possible through purely numerical studies. After giving a brief overview of our work I will focus on our current main project, the model development and analysis of a variety of column adsorption processes. The basic mathematical model of an adsorption column involves a coupled system of an advection-diffusion equation, describing the evolution of the contaminant concentration as it passes through the column, and a kinetic equation to account for the reaction process where contaminant attaches to an adsorbent material within the column. I will demonstrate how simple mathematical techniques can lead to solutions which accurately match experimental data. Along the way I will demonstrate how previous, long-accepted solutions contain errors and can be highly inaccurate, hopefully proving the power of applied mathematics!
Seminar week 9 by Jochen Einbeck
Date: 2024-11-08 at 2pm
Speaker: Jochen Einbeck (Durham Univerity)
Host: Shirin Moghaddam
Title: Statistical modelling in radiation biodosimetry
Abstract: Biological dosimetry, or short biodosimetry, refers to the task of inferring levels of radiation exposure from a retrospectively taken, potentially exposed, blood sample. When exposure to ionizing radiation occurs, this may lead to double-strand breaks which will in turn activate certain DNA-repair response mechanisms. Both the damage itself, and the ensuing repair, deliver a range of biomarkers which can be exploited in order to quantify the contracted dose. This includes dicentric chromosomes, translocations in chromosomes, micronuclei, and `foci’ from certain proteins including gamma-H2AX. These biomarkers have a simple property in common – they are all count data. As such, for the fitting of calibration curves from (designed) laboratory experiments, typical modelling choices are Poisson or multi-parameter generalized linear models for count data. However, the generation of these radiation-induced counts is subject to a variety of complex biological and physical processes, leading to phenomena such as overdispersion and zero-inflation, also sometimes requiring modelling decisions which may be deemed unusual by Statisticians. In this talk I will give an overview of the state of the art of statistical modelling in biological dosimetry, including own work especially in the context of dose estimation for the gamma-H2AX assay, and hopefully make the case that statistical dosimetry is an interesting field for Statisticians to engage with!
Seminar week 10 by Susana Gomes
Date: 2024-11-15 at 2pm
Speaker: Susana Gomes (University of Warwick)
Host: Doireann O’Kiely
Title: Control of falling liquid films using a hierarchical model approach
Abstract: The flow of a thin film down an inclined plane is a canonical setup in fluid mechanics and associated technologies, with applications such as coating, where the liquid-gas interface should ideally be flat, and heat or mass transfer, where an increase of interfacial area is desirable. In each of these applications, we would like to robustly and efficiently manipulate the flow in order to drive the dynamics to a desired interfacial shape. In this talk, I will describe a control methodology based on same fluid blowing and suction through the wall. The controls will be developed using simplified models for a falling liquid film based on reduced-order modelling and asymptotic analysis. The goal is to develop control strategies at more cost-effective levels of the hierarchy and investigate their ability to translate across the hierarchy into real-life situations by using direct numerical simulations of the Navier-Stokes equations, which in this context act as an in silico experimental framework. I will discuss distributed controls as well as (more realistic) point-actuated controls, their robustness to parameter uncertainties and validity across the hierarchy of models. If there is time, I will also discuss recent work using optimal control of similar problems
Seminar week 12 by Edouard Bonneville-Online
Date: 2024-11-29 at 2pm
Speaker: Edouard Bonneville-Online (Leiden University Medical Centre)
Host: Shirin Moghaddam
Title: Competing risks and multiple imputation of missing covariates
Abstract: When using multiple imputation (MI) to deal with missing covariates, a central consideration is that the imputation model and substantive model be compatible with each other (i.e. that they do not make conflicting assumptions). In the context of proportional hazards models in competing risks settings, a directly specified imputation model is generally only approximately compatible with the substantive model. In this talk, I will introduce various MI approaches for both cause-specific Cox models and the Fine-Gray model, and discuss how they perform based on simulation studies. I will also briefly touch upon how the development of a substantive model compatible imputation approach when using the Fine-Gray model provides insights into broader model misspecification issues in competing risks settings.
Seminar week 13 by Daniel Andre
Date: 2024-12-06 at 2pm
Speaker: Daniel Andre (Cranfield University)
Host: Cliff Nolan
Title: Multistatic Synthetic Aperture Radar Imaging and Coherence
Abstract: Synthetic Aperture Radar (SAR) imaging has proven an invaluable all-weather day and night remote sensing information source, increasingly provided by monostatic SAR satellite constellations. Future developments will include multistatic SAR constellations (with a mix of separated transmitter and receiver platforms) for more timely or simultaneous multiple collections for interferometric modes, for example for 3D imaging or more basic terrain height determination, as well as for sensitive change detection amongst others. This presentation will provide an overview of multistatic SAR processing and show highlight results from the Ground-Based SAR laboratory based at Cranfield University, which has allowed the exploration and validation of multistatic SAR sensing concepts amongst others.