Attracting and nurturing the next generation
by Meryem Abbad Andaloussi
Supervisors: Stéphane BORDAS and Andreas HUSCH
Severe brain cancers, in particular glioblastoma, have a 50% fatality rate within 15-18 months and a 5-years survival of 5%. Since 2005, the only standard of care is maximal resection surgery and concomitant radio-chemotherapy. A new research area has developed, which consists in modelling the actual biophysical processes taking place within a cancerous tumour and its surrounding. This research direction is at the crossroads between clinics, mechanics, biology and computer science. Researchers develop mathematical models which they have to tailor to the patient at hand. Setting up these models has remained challenging and costly, because no generic abstraction has emerged as capable of reproducing the specific behaviour of the general patient. What is more, a (large) number of parameters is required for those models to be practically useful, yet, limited measurement modalities have been available to identify the value of those parameters, which also evolve depending on age, medication and pathologies. For those reasons, existing modelling paradigms cannot be translated from one set of patients to another, and, thus, are severely limited in terms of their practical use. Today, a dynamic young biomechanics researcher proposes to work with two world-leading teams at the interface between the clinic (in four countries and two continents), computer science, data science and computational sciences as well as biomechanics. The team has established close collaboration with the neurosurgical department and will develop image-informed mathematical modelling based on: deep-learning image segmentation and real-time simulation, multi-scale biophysical modelling of tumour growth with uncertainty quantification, inverse modelling and error control.
The main deliverable will be the first stochastic, adaptive and error controlled open-source, open-data, open-protocol modelling and simulation approach to personalised brain cancer treatment, demonstrated in the case of meningioma. The adaptation of models to an individual patient’s morphology incorporating medical imaging data and the transportability of this modelling paradigm is our main aim. This transportability will be achieved by cross-validating models using data obtained independently of the patient set used to train the models.
Meryem ABBAD ANDALOUSSILCSBDoctoral researcher
by Mirela Puleva
Supervisors: Alexandre TKATCHENKO and Alexander SKUPIN
Computational modelling of drug-protein binding is of vital importance in the modern drug discovery pipeline as it mitigates the cost, time, and resources required to screen novel candidates for biological targets, genetic studies, and gene technology. The challenge in these investigations is two-fold: (1) even slight inaccuracies of 1 kcal/mol in the prediction of energetics in biological systems can lead to erroneous conclusions, (2) the large size of drug-protein systems means that the use of highly accurate quantum-mechanical (QM) methods is not viable. Hence, it is a grand challenge to develop efficient biomolecular force fields that can achieve and exceed stringent accuracy requirements while being efficient enough for modeling drug-protein binding in atomistic detail.
The recent combination of machine learning (ML) to accelerate QM calculations has already led to breakthroughs in our ability to obtain dynamical insights into small molecules. However, large realistic molecules remain out of reach of QM/ML approaches. In this context, the present PhD project will combine quantum mechanics, machine learning, and the theory of intermolecular interactions to develop QM/ML approaches applicable to large biomolecular systems, focusing on drug-protein binding. The main goal will be to develop a hybrid QM/ML model for capturing ubiquitous non-covalent interactions that largely determine biomolecular processes and functions.
The developed model will then be combined with existing efficient ML force fields for local chemical bonding and applied to study folding of small proteins and the interactions between drugs and their protein targets in free-energy simulations. An appropriate benchmark dataset of representative organic molecules to examine our model performance will also be generated. The aim is to reach an applicability range in system size of, e.g., the medically significant DJ-1 protein, for which our results will be compared to further calculations and experimental data from the Cell Signaling Group at the LCSB/Uni.lu.
Mirela PULEVAFSTMDoctoral researcher
by Leo Fel
Supervisors: Josip GLAURDIC and Peter RYAN
When technological progress heralded the possibility of electronic or internet voting (e/i-voting), many countries desired to introduce this novelty into electoral legislation, partly in an effort to save resources and partly in order to revitalize civic participation in the electoral process that seems to decline steadily. Over the years, the excitement has faded, and many nations with long democratic traditions have de facto blocked such innovations due to their seemingly insurmountable technological failure to reconcile demands of high assurance of the accuracy of the election result along with vote privacy while ensuring that the system is usable and understandable. Considering the continuing progress of technological advancements in this field, such a policy approach is untenable.
The experience of the US presidential elections in 2016 and 2020 and the vulnerability of e-voting to cyber-attacks further discourage the spread of e/i-voting in Europe. For the broader implementation of e/i-voting, it will be necessary to find a reasonable measure of risk that society is willing to accept, recognizing that there is a probability that technology can fail due to internal as well as external factors. This is the purpose of the proposed research project. It will bring together three disciplines – political science, law, and computing science – to find practical, political and technological solutions firmly rooted in legal theory that will enable broader implementation of e/i -voting in Europe. The Council of Europe has issued the only existing international standard on e-voting, but these standards are quite broad and reflect traditional standards clumsily applied to elections in the cyber era. Meanwhile, the world is dealing with the challenges of the COVID-19 pandemic, which has accelerated the digitalization of society and opened opportunities for a step forward in the sphere of voting technology.
The project will analyze and evaluate comparative state-level experiences in Europe, as well as the Council of Europe’s efforts to establish standards in e-voting. It will also analyze and evaluate new technological advances in e/i-voting in light of evolving legal, constitutional, and political standards to find common ground between law, politics, and technology in the 21st century. To this end, a comprehensive interdisciplinary study will be conducted that will bring together both positive and negative experiences of e/i-voting, provide an overview of current trends and provide guidelines to decision-makers to reform electoral systems in line with technological progress. The project will have a positive impact on increasing knowledge of electoral law reform, the possibilities of implementing e/i-voting in European countries, and on bringing dominant legal theories related to elections in line with technological reality and progress.
Leo FELFHSEDoctoral researcher
by Linnet Reid
Supervisors: Harlan KOFF and Conchita d’AMBROSIO
Small Island Developing States (SIDS) are particularly vulnerable to a changing climate and the constant threat of natural hazards. Gender inequalities in access to resources and opportunities systematically disadvantage women and girls, rendering them even more vulnerable to weather catastrophes, flooding, drought and rising sea-levels, and less resilient, i.e. capable of ‘bouncing back’ after a shock. While women’s disadvantageous economic situation renders them particularly vulnerable, in Caribbean SIDS, women are central to resilience building at the household and community level, given their position and role in society, as primary care givers, heads of households and entrepreneurs. Economically empowered women can champion community resilience.
A number of barriers, however limit their entrepreneurial activity in particular, including limited access to affordable and gender-sensitive financial services. In light of this reality, this project intends to investigate the potential of microfinance mechanisms to tackle gender inequalities and by extension, build resilience to climate change from the bottom up. In addition to academia, the research will be insightful for private sector actors in the financial sector, shedding light on their role in resilience building. The findings will also support policymakers formulate evidence-based solutions to development challenges in SIDS through a nexus approach. International development finance institutions stand to benefit as well from a better understanding of how to increase the effectiveness and impact of development cooperation through microfinance programmes. By studying the socioeconomic dimensions of climate change, the project is necessarily interdisciplinary in nature.
Linnet Amber Anne-Marie REIDFHSEDoctoral researcher
by Xiaowei Chen
Supervisors: Samuel GREIFF and Gabriele LENZINI
Phishing emails trick us into divulging confidential information to strangers, containing malware files or links to rogue websites and services. With the improvement of computing power and algorithms, large companies filter out most of them. However, a deal eventually ends up in the mailboxes of employees who are then left alone with the burden and the responsibility of not becoming the next victim. There is a need to improve security awareness to enhance people’s resilience, even when not all employees can be trained equally effectively, nor retain secure behaviour for long. Among the various anti-phishing training exercises, Phishing-as-a-Service (PhaaS) is most promising since the approach combines simulated phishing attacks and security awareness training.
This project studies current PhaaS solutions and develops new experimental approaches to improve their effectiveness and sustainability, inducing behavioural changes that will lead more people, and for a longer time, to reduce the click ratio of phishing emails and report suspicious emails. Specifically, controlled experiments, data analysis, HCI techniques, and longitudinal study will be applied to answer: Why are certain tactics effective for specific individual groups? How to make anti-phishing solutions more effective and sustainable? This project follows an interdisciplinary approach, leveraging expertise in cybersecurity (IRiSC research group at the SnT), knowledge of psychology and Human-computer interaction (HCI research group at the FHSE), and experience in running security training and PhaaS tools (SIU at the UL).
The research topic falls under the national research and innovation priority for Luxembourg: Industrial and Service Transformation, and is closely aligned with the “Data-Driven Innovation Strategy for the Development of a Trusted and Sustainable Economy” published in May 2019 by the Ministry of Economy to boost digital transformation of Luxembourg’s economic sector. It is aligned with the objectives of the key area “digital transformation” defined in the UL strategy framework.
Xiaowei CHENFHSEDoctoral researcher
by Nina Buntic
Supervisors: Andre SCHULZ and Jochen SCHNEIDER
At least 10 % of patients recovering from Covid-19 develop persistent health consequences such as fatigue, myalgia, or post-exertional malaise. “Long-Covid” is one of the many terms used to describe the occurrence of respiratory, cardiovascular, neurological, and/or gastrointestinal symptoms weeks after the initial infection is resolved. Although some symptoms seem to be unique for Long-Covid (e.g., olfactory & gustatory dysfunction), there is a large symptom overlap with the condition of myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). ME/CFS is a complex, multisystem condition affecting 0.89 % of the global population. Different factors have been hypothesized to be involved in the aetiology of ME/CFS, including immune system dysregulation, metabolic alteration, autonomic nervous system (ANS) and limbic system dysfunction, as well as abnormalities in the hypothalamic-pituitary-adrenal (HPA) axis. One popular hypothesis postulates that ME/CFS is a post-infectious fatigue syndrome, as up to 50 % of ME/CFS cases develop after a viral infection (e.g., infection with Epstein-Barr Virus/EBV). This observation raises the question if Long-Covid and ME/CFS share similarities in underlying pathophysiology, as both conditions seem to occur after viral infections (SARS-CoV-2 & EBV), which trigger dysregulations in the immune system, the ANS, or the HPA axis. A proper characterization of Long-Covid and ME/CFS by a thorough, interdisciplinary psychological and physiological assessment may help to make a differential diagnostic distinction of the two patient groups.
The aims of the current project are: (1) To reveal similarities and differences in the pathophysiology of Long-Covid and ME/CFS, as an in-depth understanding of the underlying psychobiology is essential to design adequate prevention in terms of early detection of pathological biomarkers and or to monitor the effectiveness of treatment interventions for Long-Covid syndromes and ME/CFS. We specifically focus on potential alterations in the immune system, the ANS, and the HPA axis. (2) We aim to elucidate how these processes translate into severity of (fatigue) symptoms, as the relationship between these alterations and actual symptom distress remains yet unclear. Ultimately, as psychobiological markers of Long-Covid and ME/CFS can help to monitor the course of symptoms and the potential responsiveness to treatment intervention, we aim (3) at investigating the effectiveness of a graded exercise therapy on symptom severity and potential improvement in alterations of the immune system, the ANS and the HPA axis.
Nina BUNTICFHSEDoctoral researcher
by Tobias Fischbach
Supervisors: Pascal BOUVRY and Alex REDINGER
The digital world is reported to consume 3-4% of the world power consumption, and increasing (9% per year), of which 20% is due to data centres [Fer18]. Data centres play a vital role in today’s cloud computing workflow and saving energy is one contribution to reduce climate change. This project aims to analyse and optimize the power demand of data centres through power and efficiency metrics derived from non-equilibrium thermodynamics. Abstraction of a data centres and even software into a network with different nodes will allow stochastic simulations of different workloads at a packet level analogous to chemical reaction networks and single-electron devices, allowing optimization of the energy. We intend to create power and efficiency metrics for several machine learning algorithms at different stages by modelling the spreading and usage of information as a thermodynamics process.
The focus on particle flow of non-equilibrium thermodynamics, prompts an analogy with the communicating processes of a distributed system. Using the information-energy equivalence in stochastic thermodynamics, we can look at improved metrics for power and efficiency. In addition to energy-efficiency, the theoretical implications can also lead to program redesign, by simplifying the program’s complexity and data requirements. The gained insights can be beneficial to the data centres (by better allocating resources to computation), but also to the programs themselves, when modelled as physical systems. Closely connected with the optimization of machine learning efficiency is the significance of data context. For humans, context allows them to dismiss data and unlearn behavior and for proper evaluation of the context in machine learning the role of dismissing data and unlearning is needed. Lastly, I will propose simple code and workshops for the communication and acceptance among the general public of machine learning algorithms and their context.
Tobias Michael FISCHBACHFSTMDoctoral researcher
by Nora Nicolai
Supervisors: Claus VOGELE and Gilbert MASSARD
Children with leukemia occurring in early childhood often experience several acute stressors as a consequence of receiving treatment, e.g., chemotherapy and chronic stressors, separation from family, which can be conceptualized as traumatic and stressful events. Previous research has shown that short- and long-term physical and mental health outcomes can be negatively impacted by early life stress (ELS). It remains unclear, however, what changes can be observed in children with early childhood leukemia at psychological, psychophysiological, molecular and epigenetic levels, and how they are related to each other. The first aim of this project, therefore, concerns the examination of psychological, psychophysiological, molecular and epigenetic processes in the context of chronic and acute stressors operationalized as leukemia in children who received chemotherapy. To prevent serious negative physical and mental health consequences of experiencing ELS in the form of chronic and acute stressors, interventions should be implemented early on for traumatized children.
Thus, the project’s second aim is to evaluate the efficacy of a brief psychological intervention to help in children with early childhood leukemia cope with adverse chemotherapy experiences. The project will be conducted as a longitudinal, clinical, prospective and multicentered study. Semi-structured interviews, self- and parent reports and questionnaires, cortisol- and blood samples are used to monitor psychological, psychophysiological, molecular and epigenetic variables. In collaboration with the Faculty of Humanities, Education and Social Sciences, the Faculty of Science, Technology and Medicine, the Luxembourg Centre for Systems Biomedicine, the Luxembourg Institute of Health and hospitals, e.g., Mutterhaus Trier, we intend to advance basic and intervention research to optimize acute treatment and aftercare concepts for children with leukemia. The expected findings will help to provide essential new insights in research on early life stress in children with severe physical illnesses.
Nora NICOLAIFHSEDoctoral researcher
by Stanislav Gubenko
Supervisors: Matthew HAPPOLD and Benteng ZOU
China is playing an increasingly significant role in global affairs: rebalancing the international political and economic order and promoting a “shift in global power to the East”. Together with China’s development reorientation, one of the most peculiar traits of this rebalancing has been China’s move from a human rights pariah state to an active participant and shaper of global human rights governance. China has successfully moved from being simply “the world’s factory” to becoming an active participant and shaper of the global and European economy, among others; advancing its economic development models and striving to build an economic system satisfying its financial and political ambitions. In the process of China’s economic norm-making and development reorientation, Europe is playing an increasingly important role. In 2013 Europe became the second largest recipient of Chinese investments, most of which came through the “One Belt, One Road” initiative. Thanks to Chinese investments, many European countries have received new development opportunities. The research seeks to analyse the interplay between mega-infrastructure projects and human rights by looking at modern Chinese infrastructure development in Europe from the human rights perspective, focusing in particular on the transformation of the Chinese approaches to human rights in foreign policy. This will be done, in particular through case studies of Italy and Ukraine: two European countries actively involved in Chinese infrastructure development.
Stanislav GUBENKOFDEFDoctoral researcher
by Alessandro Tugnetti
Supervisors: Julien PENASSE and Gilbert FRIDGEN
The importance that collectibles play in wealth management is becoming increasingly crucial, and this necessitates the development of models which can efficiently predict their price fluctuations. In this research, we will exploit the ways in which technology is changing the art market by applying supervised machine learning methods to artistic products. The objective will be to develop a pricing method that is on one hand more accurate, and on the other that maintains the level of interpretability of the models currently used in the industry. In particular, art pricing is typically concerned with estimating Ordinary Least Squares (OLS) models, leveraging on their interpretability potential.
However, it is known how these algorithms have small predictive capacity when compared with more flexible models in sectors where the uncertainty and human bias of valuation experts become stronger. Once we take advantage of the strength and abundance of data existing on the “physical” art framework, we will extend the reasoning to the Non Fungible Token (NFT) market. The goal here would be to understand the “”value-creating factors”” of NFTs, to compare them with those of classical works of art and discover points of synergy and detachment between these two worlds.
Alessandro TUGNETTIFDEFDoctoral researcher