Access to Research at National University of Ireland, Galway
Research at National University of Ireland, GalwayNot a member yet
15712 research outputs found
Sort by
Modelling fourth-order hyperelasticity in soft solids using physics informed neural networks without labelled data
No full textMild traumatic brain injury can result from shear shock wave formation in the brain in the event of a head impact like in contact sports, road traffic accidents, etc. These highly nonlinear deformations are modelled by a fourth-order Landau hyperelastic model, instead of the commonly used first or second order models like Neo-Hookean and Mooney-Rivlin models, respectively. The conventional finite element computational solvers produce robust and accurate estimates, yet they are not deployable for real-time prediction given the computational cost. The advent of physics-informed neural networks (PINNs) to solve partial differential equations (PDEs) has opened the possibility of real-time estimates of brain deformation. It involves developing a physics-informed neural network model that minimizes the residuals of the governing system of equations while ensuring boundary conditions are enforced. In this work, we propose a causal marching physics-informed neural network (CMPINN) model to capture the nonlinear mechanical response of higher-order hyperelastic materials. The CMPINN introduces a novel adaptive training scheme that incrementally updates the neural network weights. This approach incorporates several loss terms related to each material domain, boundary domain and internal domain that contributes to the total loss function, which is minimized during training. The proposed PINN framework is developed for a cube undergoing homogeneous isotropic incompressible canonical deformations: uniaxial tension/compression, simple shear, biaxial tension/compression, and pure shear. Three other tests for scenarios involving spatially varying material properties and inhomogeneous deformations are performed and benchmarked with analytical and numerical solutions.peer-reviewe
N-Terminal protein complexation and assembly with a triangular sulfated macrocycle
No full textWe report two cocrystal structures of a mutant Ralstonia solanacearum lectin (RSL) in complex with the recently described sulfated terphen[3]arene (STP3). This triangular macrocycle bearing 12 sulfates exhibits interesting protein-binding modes including methionine encapsulation and insertion between surface-exposed loops. These two binding modes facilitate the overall crystal packing, which may benefit from the macrocycle rigidity. In addition to a promising tool for protein assembly and crystallization, the data have implications for lectin–heparan sulfate interactions.We thank University of Galway, the Irish Research Council (Government of Ireland postgraduate scholarship GOIPG/2021/333 to NMM), National Natural Science Foundation of China (22001196 to MD), Natural Science Foundation of Tianjin City (23JCZDJC00660 to CL) and Research Ireland (12/RC/2275_P2) for funding. We thank SOLEIL synchrotron (Paris) for beam time allocation, and the staff at beamline PROXIMA-2A (proposal # 20210974) for assistance with data collection.peer-reviewe
A narrative review of metformin in pregnancy: Navigating benefit and uncertainty
No full textMetformin is well-established as a treatment for type 2 diabetes in non-pregnant individuals. The low cost, acceptability and broad tolerability of metformin have also made it an attractive option for research into the treatment of other conditions associated with insulin resistance. Despite almost 50 years of clinical experience with the use of metformin to treat diabetes in pregnancy, many questions remain regarding its precise effectiveness in different maternal subgroups, as well as potential short-term and long-term effects on the offspring. In this narrative review, we present the current evidence for the use of metformin during pregnancy in various maternal subgroups, including women living with overweight and obesity, women at risk of gestational diabetes, women diagnosed with gestational diabetes, and women with pregestational diabetes, including type 2 diabetes. Our specific focus is on the impact of metformin on short-term maternal, foetal and neonatal outcomes. We also consider the evidence for other emerging indications for metformin in pregnancy, such as the prevention and management of pre-eclampsia.This article was commissioned by the Editor as part of a Themed Issue on Women's health made possible by funding from Merck. Sponsoridentity was not disclosed to the authors prior to publicationpeer-reviewe
A probabilistic adversarial autoencoder for novelty detection: Leveraging lightweight design and reconstruction loss
No full textA novelty detection task involves identifying whether a data point is an outlier, given a training dataset that primarily captures the distribution of inliers. The novel class is usually absent, poorly sampled, or not well defined in the training data. A common technique for anomaly detection at present is to use an adversarial network generator to generate an anomaly score for inputs using the reconstruction loss. However, because this technique uses a competitive training process, it can be unreliable, with its performance being inconsistent during each adversarial training step. This inconsistency arises from changes in the network’s ability to detect anomalies. In this paper, we propose a revised framework for generative probabilistic novelty detection. We use a similar adversarial autoencoder-based framework but with a lightweight deep network, a novel training paradigm, and a probabilistic score to compute the reconstruction loss. Our methodology calculates the probability of whether a sample comes from the inlier distribution or not. The proposed approach can be applied to anomaly and outlier detection in images and videos. We present the results on multiple benchmark datasets, including the challenging UCSD Ped2 dataset for video anomaly detection. Our results illustrate that our proposed method learns the inlier classes and differentiates them from the outlier classes effectively, leading to better results than the baseline and state-of-the-art methods in several benchmark datasets.Research Ireland, Centre for Research Training in Artificial Intelligence, partnered with Valeo (Grant Number: 18/CRT/6223)peer-reviewe
SuPOR: A lightweight stream cipher for confidentiality and attack-resilient visual data security in IoT
No full textThe rapid growth of Internet of Things (IoT) technologies, particularly visual sensors such as cameras and drones, has resulted in increased transmission of sensitive visual data containing personally identifiable information (PII). Securing this data during storage and transmission (e.g., cloud or edge servers) is essential for maintaining privacy and security. However, existing encryption methods often face challenges due to computational overhead and vulnerability to attacks, especially on resource-limited IoT devices. To bridge this research gap, this paper presents SuPOR, a single-round lightweight cipher tailored for visual data protection in IoT environments. The SuPOR framework incorporates five fundamental cryptographic principles—Substitution, Permutation, XOR, right circular shift, and swap—which are executed in sequential steps. These include: (1) constructing a secure S-box using Möbius linear transformations and Galois fields for pixel-level substitution, (2) permuting the substituted pixels to improve diffusion, (3) applying a cryptographically secure pseudo-random number generator (CSPRNG) to generate a 64-bit one-time key for XORing, (4) performing right circular shifts on pixel byte arrays, and (5) executing element swaps to further obfuscate the data. Comprehensive security and statistical assessments demonstrate that SuPOR offers strong resistance against various attack vectors while maintaining minimal computational overhead, with a linear time complexity of
. Experimental comparisons indicate that SuPOR surpasses several state-of-the-art stream ciphers designed for IoT visual data, making it highly suitable for real-time, resource-constrained environments. The findings provide a practical and efficient solution to enhance the privacy and security of visual data in IoT systems, effectively safeguarding sensitive information from threats.peer-reviewe
Using qualitative feedback data to support educators for providing quality feedback to the undergraduate medical students
No full textIntroduction
Assessments in medical education are essential for evaluating the competencies of future healthcare professionals. Among these, Objective Structured Clinical Examinations (OSCEs) play a pivotal role by offering a structured and objective approach to evaluating clinical skills. Despite OSCEs' widespread use, significant discrepancies between observed scores and global rating scores have emerged, raising concerns about the process of reliability and validity of these assessments. These discrepancies often lead to the provision of generic, non-specific feedback, which fails to offer students actionable guidance for improvement. This thesis investigates how qualitative feedback data can better support educators in providing actionable, high-quality feedback. This exploration includes addressing discrepancies between observed scores and global rating scores, aiming to develop a feedback system that is both specific and meaningful. This study intends to empower educators to guide undergraduate medical students toward clinical proficiency.
Methods
The thesis is organised into four interrelated studies, each contributing to developing a novel structured feedback tool for OSCEs. The first study, A comparative analysis of OSCE observed scores and global rating scores using a novel approach, involved a retrospective observational analysis of scoring discrepancies between these two systems. Data were collected from 1,571 anonymised undergraduate medical students across nine cohorts. Statistical methods, including ordinal regression models and raincloud plots, were employed to identify and analyse the discrepancies between observed scores and global rating scores in OSCE assessments. The second study, A retrospective feedback analysis of objective structured clinical examination performance of undergraduate medical students, utilised text-mining techniques to analyse written feedback from 1,034 anonymised OSCE performance records. R software was used to identifycommon descriptors in the feedback, revealing a reliance on generic and non-specific terms. Thus,
the study emphasised the need for more detailed and actionable feedback. In response to the identified feedback gaps, the third study, A Systematic Review of effective quality feedback measurement tools used in Clinical Skills Assessment, systematically reviewed existing feedback measurement tools in clinical education. Databases such as PubMed, Medline, and Scopus were searched, including 14 studies. From these, ten key determinants of effective feedback—such as specificity, balance, and behavioural focus—were identified to inform the design of a new feedback tool. In the fourth study, Development and preliminary validation of a content validity index for an OSCE feedback tool in medical education , An expert panel of seven medical educators evaluated the tool's relevance and clarity across domains, including communication, task knowledge, and professionalism, and the CVI score was calculated to evaluate the structured feedback tool through the content validity index (CVI) lens, assessing its potential effectiveness in capturing and conveying essential feedback elements within the OSCE framework.
Results
The findings of this thesis highlighted significant discrepancies between observed scores and global rating scores in OSCEs, particularly in mid-range scoring categories, which emphasised the uncertainty in current assessment practices. Additionally, a retrospective analysis of the feedback provided to medical students revealed that much of it was generic, lacking the depth and specificity required to offer actionable guidance. The final studies introduced and validated an enhanced feedback tool, demonstrating its potential to address these gaps by providing medical students with more detailed, constructive, and actionable feedback that supports their clinical development.
Discussion
The identified discrepancies between observed scores and global rating scores highlight limitations of the current OSCE assessment framework. Although OSCEs remain an essential tool for evaluating clinical competencies, their effectiveness maybe may be undermined by scoring inconsistencies and generic feedback provision. These findings emphasise the need to recalibrate the feedback process, ensuring that it reflects student performance and is directive for future improvement. The introduction of a structured feedback tool offers a solution that enhances the specificity and relevance of feedback and aligns more closely with the educational goals of developing clinical proficiency in medical students.
Conclusion
This thesis stresses the need for a more structured, actionable feedback system within OSCEs. By addressing the identified discrepancies between observed scores and global rating scores and by introducing an enhanced feedback tool, this research can enhance the accuracy and relevance of feedback provided to medical students. The developed tool aims to bridge the gap between assessment and actionable feedback, ultimately improving the educational value of OSCEs and fostering the development of more competent and prepared healthcare professionals
ChiVariARIBA: a modular, editable workflow and database for characterising chitin gene variation in Vibrio spp. and related bacteria
No full textChitin is a highly abundant biopolymer of bioeconomic, biochemical and commercial importance. This carbohydrate is a source of nutrients for chitinolytic bacteria and can influence natural competence, surface adsorption and other fundamental aspects of prokaryote physiology. Bacterial enzymatic degradation of chitin is mediated by a well-studied set of hydrolytic enzymes, transcriptional regulators and carbohydrate transport proteins. Many of these gene products have been functionally characterized in vitro or in vivo, but there is a reliance on in silico genomic approaches to study the variation of these metabolic components amongst diverse bacteria. Computational surveys of bacterial genomes to date have tended to focus on determining the presence and absence of chitin metabolism genes in diverse genomes, but not on the diversity of sequences amongst these gene families. To enable future research into chitin metabolism variation in vibrios and other bacteria, we present ChiVariARIBA, a workflow for extracting chitin metabolism genes from published genome sequences of chitinolytic Vibrio species and their relatives, compatible with the rapid gene-finding and variant-characterizing tool ARIBA, with which to describe the presence of chitin-metabolising genes in genomes of interest and to characterize the sequence variation of these genes across diverse bacteria.E.P.N. was funded by a strategic research award from the University of Galway’s College of Science and Engineering Sustainable Development Goals Research Support Fund (awarded to M.J.D. and which supported the work described in this manuscript). For the purposes of Open Access, the authors have applied a CC-BY public copyright licence to any author-accepted manuscript version arising from this submission.peer-reviewe
Counter-speech generation for homophobic and transphobic social media content in Malayalam
No full textThe growing prevalence of hate speech online has amplified acts of discrimination against marginalized populations, with the LGBTQIA+ community being particularly affected. In areas where under-resourced languages such as Malayalam are used, the issue grows more complex because of the absence of localized resources. This research offers an in-depth analysis of the production of counter-speech to address transphobia and homophobia in Malayalam. Our work incorporates both native Malayalam script and Malayalam written in Latin script, addressing the diverse linguistic practices of online users in Kerala. This paper introduces a two-stage pipeline to counter such online abuse. The first stage focuses on dataset creation through a human-in-the-loop process, beginning with 100 seed pairs of hate speech and their corresponding counter-speech manually curated. This set is expanded iteratively using language models culminating in 5,000 validated pairs. In the second stage, we propose a method to generate counter speech in Malayalam that leverages the Retrieval-Augmented Generation framework enhanced by REFINE (Retrieval Evaluation via Fluency, Inversion, and NEarness) for knowledge retrieval and constrained decoding. Evaluation metrics for both dataset quality and model performance demonstrate the effectiveness of this approach in producing diverse, fluent, and target-specific counter-speech. This research provides a foundational resource and scalable strategy for countering hate in low-resource regional languages. GitHub Link: https://github.com/Bharathi-AI-for-Social-Good/CN-Malayalam.Open Access funding provided by the IReL Consortium. This research has not been funded by any company or organization.peer-reviewe
Performance optimization of an office building with a dynamic façade system
No full textA calibrated building energy simulation model assessed a dynamic façade system (including shading and natural ventilation) in an office, achieving a mean absolute error of 2.1% for heating and cooling. Several operational strategies were evaluated, optimizing shading via solar irradiance setpoint. Integrating natural ventilation yielded a 14.9% energy reduction compared to a no-ventilation scenario. Sensitivity analysis showed minimal total energy changes with different occupancy densities and natural ventilation rates, yet occupancy density strongly affected the heating-to-cooling ratio. Broadening heating and cooling setpoints further reduced energy use, highlighting the system’s significant potential for efficient office design.This work was supported by the European Union’s MSCA PF FaceINQ project under Grant number 101066362; the Academic Committee for Design of the Department of Mechanical Engineering, Eindhoven University of Technology under Grant number OA102424-10.peer-reviewe
Simulation-driven design and optimisation tools for additive manufacturing applications
No full textAdditive manufacturing technologies such as laser powder bed fusion have been widely adopted in various industries. A core reason for this adoption is the design freedom which allows for more complex designs than conventional manufacturing methods. While design for additive manufacturing includes restrictive aspects to ensure parts are manufactured successfully and at high quality, there is also an opportunistic aspect where the designer can improve performance by consolidating assemblies, reducing weight, and improving performance with new fabrication capabilities. Engineering design and analysis tools have also advanced greatly, with simulation being a core part of engineering design and iteration. To take full advantage of these new additive design freedoms new simulation-driven-design tools are required which optimise complex designs based on state-of-the-art in-service simulations and for additive manufacturing.
This thesis develops and demonstrates a simulation-driven-design framework which automates the geometry generation, simulation and design iteration/optimisation steps to speed up design iteration and optimise part design. Knowledge of additive manufacturing limits are encoded in the framework to ensure the final designs are manufacturable and of acceptable quality. This is applied to develop and demonstrate three problem-specific design tools which advance the state of the art in the demonstrator topics: lattice structures, conformal cooling for injection moulding, and thermal stability of space optics.
First an inverse design tool is developed for lattice structures which designs lattices to exhibit a target compressive stress-strain curve based on beam-element finite element analysis and accounts for large deformations and contact. This is supported by validation of the model with experimental testing with an average error in plateau stress of 12.6%, and X-ray micro-computed tomography to measure lattice dimensions. The tool is applied to design lattices with five distinct stress-strain curves, which are manufactured, tested, and compared to the target curves, achieving errors in plateau stress of 4.5–14.5% for four of the five test cases. Lattice structures are studied further via multi-material powder bed fusion, exploring another dimension of design freedom offered by the latest developments in multi-material additive manufacturing. Bi-metallic 17-4PH/316L stainless steel lattices are manufactured and the material interface is studied in lattice and bulk specimens. The material interface was found to be robust under the compressive loading, with no failure or cracking initiating at the interface. The compressive behaviour of two contrasting unit cells and two loading orientations are studied, with bi-metallic samples found have a greater energy absorption than single-material samples. A finite element model is developed to predict the response, including as-built dimensions informed by computed tomography. Finite element model predictions of the first maximum strength differ by a range of 3.2–96.6% to experimental values.
To improve cooling time and cooling uniformity in injection moulding, an automated design tool is created to automatically design and optimise conformal cooling channels. The tool is demonstrated in a generic case study as well an industrial problem to design cooling channels for a real mould tool featuring multiple inserts. Channels are designed in a two-step process where first a candidate design is selected from a parameter study, and this candidate proceeds for further optimisation to improve the temperature uniformity. Analysis is performed with a MATLAB-based thermal model which is compared to industry-standard software Moldflow. In both demonstrators, the temperature uniformity and cooling time are improved via the MATLAB-based optimisation, however due to some of the assumptions in the MATLAB model improvements in temperature uniformity do not carry over to Moldflow in scenarios where some channels increase in temperature up more than others.
To improve the thermal stability of optics in space applications, a custom shape optimisation method is developed and applied to minimise the optical wavefront error resulting from thermal gradients. An initial design is first created with topology optimisation to optimise the structural performance and find a lightweight design, which is directly followed by a custom shape optimisation step to optimise the optical performance measured based on a ray tracing simulation before and after thermally-induced deformations. This shape optimisation step reduced the wavefront error by up to 79%, and multi-objective optimisation resulted in simultaneous reductions in wavefront error by 60% and in volume by 20%. This custom method allows for further optimisation directly on the output of topology optimisation, and enables shape optimisation to improve on optical performance under a prescribed thermal load.
This project has demonstrated the design framework across structural, thermal, and coupled thermomechanical and optical problems, producing a series of design tools offering fast design iteration and optimisation for additive manufacturing applications. This work has also produced a number of findings for design and design methods for each topic as a result of the large batches of simulations that are run with this method, showing its value as a research tool as well as for simulation-driven-design