263134 research outputs found
Sort by
The twisted conjugacy growth of virtually abelian groups
No full textsponsorship: We would like to thank the anonymous referee for their valuable remarks. Karel Dekimpe is supported by Methusalem grant METH/21/03-long term structural funding of the Flemish Government. Maarten Lathouwers is funded by FWO PhD-fellowship fundamental research (file number: 1102424N). (Flemish Government|METH/21/03, FWO PhD-fellowship fundamental research|1102424N)status: Publishe
Van het ongrijpbare naar het meetbare: Machine learning voor het modelleren van prestatiebepalende factoren in de sport
No full textOver the past two decades, sports analytics has emerged as a powerful tool for advancing our understanding of athletic performance and ultimately gaining a competitive advantage in sports. The use of wearable sensors, video tracking, and other monitoring technologies has enabled large-scale data collection. When combined with statistical analysis, these data have transformed how teams and athletes approach training, competition, and strategic decision-making. Data-driven insights now underpin talent identification, opponent scouting, injury prevention, and load management.
Despite these advancements, certain aspects of sports performance remain difficult to quantify. Many of the qualities practitioners seek to understand—such as leadership, teamwork, and decision-making under pressure—are inherently intangible and cannot be directly measured through objective data points. Other parameters may be too costly or impractical to assess regularly. For example, in endurance sports, laboratory tests such as VO₂max protocols and force-plate assessments provide gold-standard measurements but are impractical for frequent, in-field use.
This dissertation explores how machine learning can provide value to the analysis and interpretation of these elusive aspects of performance. The work focuses on two specific domains: (1) the analysis of biomechanical parameters in outdoor running, and (2) the assessment of in-game decision-making in professional soccer.
In running, monitoring biomechanical load is essential for injury prevention, but obtaining such data typically requires expensive, lab-based equipment. In this dissertation, we explore how machine learning can provide a scalable alternative by estimating key biomechanical parameters from wearable sensor data. Specifically, we use bilateral shin-mounted accelerometers to estimate (1) gait events to derive ground contact time, and (2) vertical instantaneous loading rate (VILR), a kinetic variable linked to injury risk. Our results demonstrate that machine learning models can accurately estimate these parameters during overground running and outperform heuristic methods, offering a promising step toward real-time, field-based feedback systems for injury monitoring and prevention.
In soccer, decision-making is central to performance, yet data only captures the action taken—not the alternatives a player could have considered—making it difficult to assess whether a player made the optimal choice in a given situation. This dissertation approaches decision-making from three complementary angles. First, we examine how context influences decisions by modeling both pre-game and in-game factors to estimate mental pressure, allowing us to assess how players respond under varying psychological demands. Second, we explore creativity as a specific aspect of decision-making, identifying effective passes that deviate from typical choices. Third, we focus on pressing as a tactic that requires coordinated decision-making and propose a framework to quantify the impact of pressing in terms of how a decision to press affects the long-term risk and reward. Together, these frameworks offer a data-driven, objective lens on decision-making in professional soccer.
Together, these case studies illustrate the promises and limitations of machine learning to assess complex, difficult-to-measure aspects of sports performance. While our models deliver actionable insights, their validity hinges on careful feature design, critical evaluation, and close collaboration with coaches and sport scientists. Future research should prioritize athlete- and team-specific modeling to capture individual style and tactics, and integrate richer contextual variables such as fatigue, opposition behavior, and environmental factors. Advances in interpretable modeling and uncertainty quantification will be essential to foster practitioner trust and ensure that analytics guide, rather than oversimplify, decision-making in high-stakes sporting environments.status: Publishe
To punish or to assist? Divergent reactions to ingroup and outgroup members disobeying social distancing
No full textstatus: Publishe
Impact van vertragingen bij de diagnose en korte behandelingsregimes op de resultaten bij patiënten met rifampicine-resistente tuberculose in Kameroen
No full textSupervisors:
Prof. dr. Tom Decroo (ITM)
Prof. dr. Emmanuel André (KU Leuven/UZ Leuven)
Prof. Palmer Masumbe Netongo (Université of Yaoundé I, Cameroon)
Prof. dr. em. baroness Lut Lynen (former director ITM)
Abstract:
My thesis aimed to identify the gaps in RR-TB diagnosis and treatment in Cameroon and suggest areas for improvement. In the first objective, using spatial analysis and mathematical modelling, I identified regions with probable low TB and RR-TB notification, including the corrected national TB and RR-TB diagnostic gap for each of the ten regions. Considering that at present, 58% and 28% of WHO-estimated TB and RR-TB patients are being diagnosed, this result could support the efficient allocation of available rapid molecular diagnostic tools nationwide. This may greatly impact closing both TB and RR-TB diagnostic gaps.
In the second objective, using a systematic review and meta-analysis, I showed that in high TB/HIV burden countries(including Cameroon), after more than a decade of implementation of the Xpert MTB/RIF as point-of-care TB diagnosis, up to 18% (95%CI: 12-25)% of diagnosed RR-TB patients do not start treatment (most studies being routine program data) while reasons were not systematically reported. Therefore, it will be essential in Cameroon to routinely report pre-treatment attrition in all patients diagnosed with RR-TB, along with the reasons for not starting treatment, to tailor future interventions aimed at closing the RR-TB pre-treatment gap.
Finally, in the last objective, using nationwide data over five years, I demonstrated the importance of addressing missing second-line DST data (29.6% in this cohort) for key TB drugs (quinolones and injectables) used to manage RR-TB as a key predictor of mortality. This should definitely be a priority when constructing effective regimens, including using bedaquiline, to sustain high treatment success while safeguarding resistance acquisition to the core drugs used.
In summary, my work provides efficient methods for improving the RR-TB diagnostic and therapeutic pathway in Cameroon. More efficient use of diagnostic means may substantially increase RR-TB notification. Identifying the RR-TB pre-treatment gap and its determinants will make the RR-TB programme more effective. Access to DST for the most important second-line drugs will improve treatment outcomes. Altogether, these interventions will contribute to better RR-TB control in Cameroon.status: Publishe
Onderzoek naar moleculaire nanotechnologie voor de detectie van individuele moleculen met vaste-stof nanoporiën
No full textNanopores enable label-free single-molecule sensing with broad application potential in genomics, proteomics, drug discovery, and molecular data storage. Their exceptional sensitivity is exemplified by the successful commercialization of DNA sequencing using biological nanopores. This dissertation focuses on solid-state nanopores, which offer a distinct set of advantages over biological nanopores, including mechanical robustness, scalability through dense nanopore arrays, improved noise characteristics, and compatibility with CMOS integration for high-bandwidth measurements.
Despite these benefits, fabricating solid-state nanopores with atomic-scale precision and sub-5 nm dimensions remains extremely challenging. This size mismatch between solid-state nanopores and the fine structural features of most biomolecular analytes limits their sensitivity.
To overcome these limitations, this dissertation introduces novel strategies inspired by nature, using molecular nanotechnology to enhance single-molecule detection with solid-state nanopores. Specifically, the predictable folding of DNA nanostructures and the self-assembly of lipid entities are harnessed to design structural building blocks and functional components.
Two central research questions are explored: (1) What are the minimal structural requirements for reliable detection of DNA nanostructures using solid-state nanopores? (2) Can programmable biomolecular components be used to engineer scalable hybrid nanopore systems?
In the first part, DNA nanostructures were designed with large features, or labels, that match solid-state nanopore dimensions, enabling high-sensitivity readout. This approach addresses a fundamental limitation: solid-state nanopores are not yet capable of resolving the fine structural details of native DNA. The engineered DNA nanostructures with large labels, referred to as 'DNA structural barcodes', offer a direct route to scalable biosensing by avoiding the need for complex nanopore geometry tuning. This work presents the first systematic investigation of the minimum size requirements for barcode labels in membrane-based solid-state nanopore systems. Chapter 2 reveals that the physical characteristics of barcode labels influence their local translocation velocity, distinct from the DNA backbone, and significantly affect readout success.
In the second part, two hybrid nanopore concepts are introduced to combine the strengths of biological and solid-state nanopores. The integration of the highly sensitive biological nanopores into solid-state platforms may ultimately enable the readout of fine structural features in biopolymers, like DNA. Existing hybrid strategies face technical challenges such as leakage, instability, and poor integration. To address these, programmable DNA and self-assembling lipids are used to develop engineerable components for hybrid nanopore formation.
The first concept presents a DNA-lipid interposer that stabilizes biological nanopores in a lipophilic environment and enables guided docking onto solid-state nanopores. Chapters 3 and 4 detail the development and assembly of this interposer, resulting in the first successful integration of a lipid bicelle into a DNA origami structure. This novel DNA-lipid nanodisc marks a significant advancement in nanoscale membrane engineering. It provides a versatile platform for single-molecule studies of membrane proteins and was realized by two methodological innovations not previously reported: a detergent-free bicelle incorporation protocol and the use of DNA origami as a structural stabilizer. These approaches allow robust and tunable membrane integration within DNA nanostructures. Additionally, experimental evidence of biological nanopore monomer insertion represents a promising step toward realizing the envisioned hybrid nanopore platform.
The second concept demonstrates, for the first time, the formation of pore-spanning supported lipid bilayers through the fusion of nanoscale lipid discs (i.e., bicelles). Chapter 5 establishes this method as a scalable and robust approach for hybrid nanopore assembly. Compared to liposome-based techniques, bicelles offer a more favorable geometry for spanning nanopores and enable high sealing resistance, which is essential for achieving high sensitivity. This streamlined strategy supports the development of high-density hybrid nanopore arrays with minimal fabrication complexity.
Together, these contributions advance solid-state nanopore technology by introducing scalable and nature-inspired molecular strategies for improving the sensitivity. The use of DNA structural barcodes and hybrid nanopore platforms opens new possibilities for single-molecule biosensing and membrane protein studies in native-like environments.status: Publishe
One-year effectiveness and safety in young children aged 2-6 years with type 1 diabetes using an automated insulin delivery system: A real-world prospective cohort study
No full textAIMS: This study evaluated 1-year real-world changes in glycaemic management, parent-reported outcomes and safety with an automated insulin delivery (AID) system in young children with type 1 diabetes (T1D). MATERIALS AND METHODS: Children aged 2-6 years whose parents agreed to initiate the Medtronic MiniMed™ 780G were enrolled at 15 centres between October 2022 and December 2023. Data were collected quarterly over 1 year during routine follow-up. Parent-reported outcomes were assessed using questionnaires (HAPPI-D [part of the HAPPI-D Protocol; Hvidøre, Adolescent, Parent, Professional, Instrument, Diabetes] and Hypoglycaemia Fear Survey [HFS]-Parent). The primary endpoint was the evolution of time in range (TIR, 70-180 mg/dL) from start to 12 months after start. Data are reported as mean ± SD or least-squares mean (95% confidence interval). RESULTS: A total of 149 children were included (mean age 4.2 ± 1.4 years; 56.4% girls). Mean T1D duration was 22.0 ± 13.0 months and 75.2% used an insulin pump before. After 1 year, TIR increased from 56.8% (54.4-59.2) to 66.6% (64.7-68.5) and haemoglobin A1c decreased from 7.6% (7.4-7.8) to 7.2% (7.1-7.4) (all p < 0.001). Time <70 mg/dL remained stable (5.0% [4.2-5.8] at start vs. 4.6% [3.9-5.3] at 12 months, p = 0.172). Parents reported less diabetes burden on the HAPPI-D (22.9 points [21.7-24.0] at start vs. 21.7 points [20.5-22.8] at 12 months, p = 0.001), while scores on the HFS-Parent did not change significantly. There were no hospitalisations for severe hypoglycaemic events and one for diabetic ketoacidosis due to infusion set occlusion. CONCLUSIONS: One-year use of an AID system in young children with T1D was safe and associated with improved glycaemic management and reduced parental burden, with limited impact on time in hypoglycaemia and related parental fear.sponsorship: Medtronic (Medtronic)status: Accepte
Een multischalig computationeel kader toegepast op de pyrometallurgische recyclage van platinumgroepmetalen
No full textSince the 1970s, automotive catalysts have been used to regulate air pollution resulting from incomplete fuel combustion. Their active components are Platinum Group Metals (PGM), more specifically Pt, Pd, and Rh, typically present as nanoparticles dispersed within a porous γ-Al2O3 washcoat layer mounted on a cordierite (2MgO-2Al2O3-5SiO2) substrate. At the end of their service life, efficient recycling of PGMs from spent automotive catalysts (SACs) is essential for both sustainability and economic reasons.
Pyrometallurgical smelting is the most widely applied industrial recycling method, used by companies with advanced metallurgical and refining technologies, including Umicore (Belgium). In this process, ground SACs are melted with fluxes (mostly CaO), a collector metal (typically Cu for cordierite-based catalysts), and a reducing agent. During the smelting process, a slag phase forms from the molten cordierite and flux, while dispersed Cu droplets capture the PGMs, leaving the Al2O3-carrier in the slag. Despite its industrial maturity, the atomic and microscale mechanisms controlling PGM collection remain poorly understood.
This thesis develops a comprehensive, multi-scale modeling framework to quantitatively simulate the recovery of PGM nanoparticles by Cu droplets within a slag environment. The approach integrates a thermodynamically consistent, multi-phase, multi-component phase-field model with kinetic parameters obtained from molecular dynamics (MD) simulations.
The first part of this work extends conventional phase-field models, which often encounter limitations under large thermodynamic driving forces or in the presence of stoichiometric compounds. By rederiving the governing equations for stoichiometric compounds and generalizing the high-driving-force adaptation to multi-component systems, a robust framework was established and validated. This enables stable, quantitative simulations of diffusion-controlled transformations across all driving forces and length scales.
The second part addresses the scarcity of kinetic data for oxide melts, a key limitation for quantitatively simulating Al2O3 washcoat dissolution in CaO-Al2O3-SiO2 slags. A high-throughput MD study quantified the composition- and temperature-dependence of Ca, Al, Si, and O self-diffusion in these melts. Bouhadja's empirical force field was identified as most accurate for CaO-Al2O3-SiO2 melts through benchmarking against experiments, empirical models, and ab initio MD for the structural and transport properties of the melt. Using this force field, 119 simulations were performed across 40 compositions and five temperatures. The resulting diffusivities were parameterized using the pair-fraction-based model of Thibodeau and Jung. Both MD-only and hybrid (MD+experiment) fits were successfully validated against diffusion couples and dissolution experiments.
Finally, the integrated framework was applied to simulate key PGM recovery mechanisms. Simulations showed that PGM dissolution in liquid Cu occurs almost instantaneously once contact is established, whereas the release of encapsulated PGMs from the wash coat is several orders of magnitude slower, making it the rate-limiting step in early smelting. The results highlight slag design as a critical lever for accelerating washcoat dissolution, while maximizing collision probability, through optimized collector content and slag compositions that promote effective mixing, emerges as the most impactful strategy for recovery efficiency.
This thesis demonstrates how phase-field modeling and molecular dynamics can be tightly coupled to quantitatively predict the fundamental small-scale mechanisms underpinning PGM recovery in smelting conditions. The combination of methodological advances, including an extended phase-field framework and an MD workflow to obtain reliable kinetic parameters for oxide melts, with systematic application to Cu-based PGM collection, provides both new insight into industrially relevant processes and a transferable modeling strategy. By linking atomic-scale transport with mesoscale microstructure evolution, this work establishes a foundation for predictive simulations in metallurgical systems where complex chemistries and multi-phase interactions govern performance.status: Publishe
Hoge frequentie resonatoren om supergeleidende ringen te onderzoeken. Van metastabiele toestanden tot kwantumcoherentie
No full textSuperconducting circuits combine zero electrical resistance with macroscopic quantum coherence, making them a versatile platform for both fundamental studies and technological applications. Modern thin-film deposition and lithography allow precise control over geometry, material composition, and dimensional scales, enabling the fabrication of complex superconducting structures with tailored properties. These capabilities give access to phenomena not achievable with conventional electronic components, establishing superconducting circuits as a unique tool for exploring quantum physics and developing devices with unique functionalities.
In superconducting rings, macroscopic quantum coherence manifests as discrete circulating currents. Fluxoid quantization restricts these currents to discrete winding states. Transitions between winding states occur via phase-slip events. A phase slip event requires overcoming an energy barrier, which can be crossed either through thermal activation, resulting in thermally activated phase slips (TAPS), or via quantum tunneling, resulting in quantum phase slips (QPS) that have no classical analogue. Both processes depend sensitively on temperature, loop geometry, material properties, and the intrinsic properties of the superconductor. Embedding weak links or nanowires in a superconducting ring provides localized regions where phase slips preferentially occur, enabling controlled TAPS and QPS dynamics. When the QPS rate becomes sufficiently high, coherent tunneling between neighboring winding states generates an energy splitting, forming the basis of qubit states. The discrete winding states can thus be exploited for device applications: serving as robust memory elements when the energy barrier is high and phase slips are rare and well controlled, and as qubits when coherent quantum phase slips allow the system to occupy superpositions of distinct winding states, marking the transition from a classical bit to a quantum bit.
Phase-slip dynamics in superconducting rings have traditionally been studied using switching-current measurements, in which the bias current is ramped until the ring or nanowire becomes resistive. While this approach provides statistical information about phase-slip rates, it is inherently dissipative and cannot detect events that do not drive the system in the normal state, leaving many phase slips hidden and preventing their unambiguous classification by origin or rate.This thesis develops a non-invasive, radio-frequency resonator-based platform for studying phase-slip dynamics in superconducting rings. Building on magnetooptical imaging of flux penetration in superconducting resonators, we first established methods to apply a local magnetic field to a loop without perturbing the resonator and couple it to the resonator, linking changes in circulating currents to shifts in the resonance frequency. This enabled continuous, timeresolved detection of individual phase-slip events on microsecond-to-nanosecond timescales. Using this platform, we systematically investigated how loop geometry, weak-link properties, kinetic inductance, and temperature govern phase-slip rates and dynamics. Deterministic control of discrete winding states in high-kinetic-inductance aluminum loops, our chosen material platform, was demonstrated, realizing a superconducting memory based on fluxoid quantization. By progressively reducing loop dimensions and introducing weak links, we observed quantum phase-slip events at millikelvin temperatures,
providing the first experimental signatures of coherent tunneling between winding states, a first step toward realizing phase-slip-based qubits.
Collectively, these results establish a robust experimental pathway from conventional DC measurements to non-invasive RF probing, provide a framework for controlled studies of superconducting phase-slip dynamics, and open the door to exploring and utilizing coherent phase slips in circuit-QED-compatible architectures with device geometries that are simple and easy to fabricate.status: Accepte
Het exploreren van nieuwe toepassingen van peripartale echografie.
No full textThis thesis aimed to explore new applications of ultrasound use during labor and delivery to predict and ultimately improve maternal and neonatal outcomes.
First, we focused on implementing intrapartum ultrasound (IPUS) in our labor suite and explored how artificial intelligence can aid this process. In Chapter 3, we demonstrated that the sonographic assessment of fetal head position and station requires minimal theoretical knowledge, which can be easily acquired in a short period. Midwives, however, indicated low confidence levels for performing IPUS themselves, most notably for the sonographic assessment of fetal head station. Therefore, in Chapter 4, we in-house developed and validated an artificial intelligence (AI) algorithm for automated assessment of the angle of progression from transperineal ultrasound (TPUS) volumes acquired in labor. Computer-observer angular differences were within expert inter-observer variation, demonstrating excellent performance of this proof-of-concept AI model.
Next, we investigated whether ultrasound could predict certain adverse labor outcomes: unplanned operative delivery (Chapter 5), early adverse perinatal outcome (Chapter 7), and postpartum anal incontinence (Chapter 9). Before these clinical studies, we performed comprehensive literature reviews to identify knowledge gaps and inform the design of these studies (Chapter 6 and Chapter 8). We hypothesized that study-specific ultrasound parameters could improve screening performance. A similar statistical analysis strategy was adopted in each study. For each outcome, a clinical prediction model was fitted, including history-based covariates that were known at the time of admission to the labor ward. Outcome-specific sonographic predictors were subsequently added to these baseline models to assess changes in screening performance. In Chapter 5, the addition of levator ani hiatal biometry and fetal cerebroplacental ratio significantly improved screening performance for unplanned operative delivery in a cohort of nulliparous women admitted for spontaneous or induced labor at term. The latter was not observed in Chapter 7, where the addition of Doppler sonography findings or estimated fetal weight did not improve screening performance for operative delivery for suspected fetal compromise or adverse perinatal outcome in a mixed parity cohort. In Chapter 9, the sonographic subpubic area, a novel estimator of maternal pelvic outlet dimensions, was an independent predictor of postpartum anal incontinence. However, it did not improve the identification of nulliparous women at risk of developing postpartum anal incontinence after their first vaginal delivery.status: Publishe