In this thesis, deep learning is studied from a statistical perspective. Convergence rates for the worst case risk bounds of neural network estimators are obtained in the classification, density... Show moreIn this thesis, deep learning is studied from a statistical perspective. Convergence rates for the worst case risk bounds of neural network estimators are obtained in the classification, density estimation and linear regression model. Special attention is given to the role of input dimension since in practice, neural networks have shown promising results for high dimensional input settings. First, the estimation of conditional class probabilities under the cross-entropy loss is studied. A challenge with this loss is that it becomes unbounded near zero. To deal with this, the loss is truncated. Convergence rates are obtained for a neural network estimator under this truncated loss. The second problem considered is density estimation. A two step procedure is proposed. The first step transforms the density estimation problem into a regression problem by constructing response variables using a kernel density estimator on half of the data. In the second step, a neural network is fitted to this constructed data. Convergence rates for this method are obtained using existing approximation results for compositional functions. Finally, forward gradient descent is studied. This is a biologically motivated alternative for gradient descent. Convergence rates are derived for this method in the linear regression model with random design. Show less
The studies described in this thesis provide additional evidence that the current Dutch population-based screening programmes (CSPs) can be further optimized, particularly concerning the... Show moreThe studies described in this thesis provide additional evidence that the current Dutch population-based screening programmes (CSPs) can be further optimized, particularly concerning the participation of potential participants from highly urbanized and low socioeconomic status (SES) neighbourhoods. Our findings suggest that non-participation in the CSPs in these low SES-neighbourhoods is associated with more unfavourable, relatively advanced, tumour outcomes. Given that the decision to participate in a CSP is not solely based on rational decision-making processes, primary care providers could play an important role here. This would primarily involve informing and advising potential participants who are hesitant about participating in CSPs. In this thesis, we describe that both potential participants and general practitioners support the idea that primary care should be more involved in the invitation process of the CSPs. Based on our findings, we therefore recommend implementing a proactive, risk-based invitation strategy from primary care regarding the invitation process of the current CSPs. Show less
This dissertation examines various monitoring mechanisms within the International Humanitarian and Human Rights Law frameworks in order to draw lessons learned from their monitoring functions,... Show moreThis dissertation examines various monitoring mechanisms within the International Humanitarian and Human Rights Law frameworks in order to draw lessons learned from their monitoring functions, approaches and practices and assess whether they can be applied to a future compliance system under the four Geneva Conventions of 1949 and their Additional Protocols. The scope of the enquiry includes the existing compliance system of the Geneva Conventions, the compliance systems of other treaties within the International Humanitarian Law framework, the United Nations human rights monitoring system, and a number of mechanisms that supervise respect for humanitarian norms by States and/or armed non-State actors. The dissertation suggests that a future compliance system under the Geneva Conventions should include thematic reporting, a plenary forum for States Parties and an International Humanitarian Law Committee. The establishment of effective compliance mechanisms that are tailored to the specifics of International Humanitarian Law would support and advance the implementation of the Geneva Conventions by States Parties as well as strengthen the capacity to monitor their implementation from within the International Humanitarian Law framework. Show less
This thesis had two main focuses. First, it investigated the mechanisms underlying individual differences in brain structure, developmental changes, and function. Second, it examined how enriched... Show moreThis thesis had two main focuses. First, it investigated the mechanisms underlying individual differences in brain structure, developmental changes, and function. Second, it examined how enriched and deprived environments affect brain development from middle childhood to early adolescence (ages 7-14). The study revealed distinct genetic and environmental influences on the developing brain during this period. Genetic contributions largely explained the variations in brain structure and developmental changes, along with location-specific environmental factors. While studies on genetic and environmental influences on brain regions showed both overlapping and unique outcomes, the interplay between these influences is complex. Using MRI and fMRI provided a comprehensive understanding of how individual differences arise. Additionally, the thesis highlighted the distinct effects of specific environmental influences on brain development. An enriched cognitive environment, such as musical ability, was linked to attenuated development, while a deprived environment, exemplified by the COVID-19 pandemic, accelerated development. Notably, the brain showed resilience to adverse effects, particularly in TPJ growth during the pandemic, suggesting adaptive processes during this developmental phase. Future research should explore whether the relationship between environmental effects and brain maturation pace is linked to positive mental health outcomes. Ultimately, identifying interventions to ensure all children thrive throughout their development is crucial. Show less
It is increasingly common for citizens to take initiatives to improve their immediate living environment. While setting up and implementing these initiatives initiators come into contact with the... Show moreIt is increasingly common for citizens to take initiatives to improve their immediate living environment. While setting up and implementing these initiatives initiators come into contact with the government. Governments are often enthusiastic about these plans, but yet it proves far from easy to get get the initiative off the ground. 'The law' is seen as the main bottleneck. In this dissertation several (legal) bottlenecks are researched and recommendations to facilitate initiators are formulated. Show less
If diversity counteracts solidarity, as welfare state scholarship commonly concludes, then how did the expansion of European welfare states in the postwar period coexist with (post)colonial... Show moreIf diversity counteracts solidarity, as welfare state scholarship commonly concludes, then how did the expansion of European welfare states in the postwar period coexist with (post)colonial migrations? Through a historical-interpretivist lens, this dissertation studies the inclusion of migrants from the Caribbean, Algeria, and present-day Indonesia in post-war British, French, and Dutch welfare setates respectively. It documents a variety of inclusions on inequal terms in French and Dutch welfare states. While those who qualified as repatriates received aid under targeted assistance schemes and expedited access to social security, many who had been disadvantaged under colonial legal codes were directed towards paternalistic arms of the welfare state aimed at surveillance and cultural conversion. In the UK, street-level discrimination and restricitonist immigration law marred the promise of equality under the Beveridgean welfare state for Caribbean citizens. These patterns were racialised, but not inevitable. Ideological efforts to construct certain migrants as deserving cultural insiders have important explanatory power. This dissertation improves the theoretical toolkit available to welfare state scholars interested in inclusion, and elucidates the relationship between race and redistribution. Diversity neither caused retrenchment, nor threatened solidarity. Rather, (post)colonial migrations offered contemporaries an opportunity to consciously and manually create it. Show less
Oratie uitgesproken door Prof.dr. Ben S. Kuipers bij de aanvaarding van het ambt van hoogleraar Publiek Leiderschap aan de Universiteit Leiden op vrijdag 14 juni 2024
Lemkes, A.C.; Struijs, S.Y.; Littel, M.; Ekkers, W.D.; Berg, J.F. van den 2024
This research focuses on creating composite biomarkers that can classify diagnoses, estimate symptom severity, and detect treatment effects using data from wearable sensors and smartphone... Show moreThis research focuses on creating composite biomarkers that can classify diagnoses, estimate symptom severity, and detect treatment effects using data from wearable sensors and smartphone applications. The thesis consists of an introduction to machine learning techniques and their use in developing biomarkers for the central nervous system; a narrative review of the relevant literature; and detailed studies on the application of these techniques in various health conditions. Specifically, the research includes observational and cross-sectional studies on facioscapulohumeral muscular dystrophy (FSHD) and major depressive disorder (MDD), demonstrating how smartphone and wearable sensor data can be used to monitor disease severity and progression. Additionally, the research identified the use of a tablet-based finger tapping task to monitor the real-time effects of antiparkinson's drugs on Parkinson's symptom severity. Key findings highlight the potential of mHealth biomarkers to provide continuous, real-time monitoring of patients, which can enhance the accuracy of clinical assessments and potentially reduce the burden on patients and healthcare systems. The thesis also addresses the challenges of variability in mHealth device data and emphasizes the need for robust validation and standardization to ensure the reliability of these biomarkers in clinical settings. Show less
The human body consists of many different cell types. Cell types can be defined by the genes expressed, and unique cell-type-specific transcriptional mechanisms control these expressions. Single... Show moreThe human body consists of many different cell types. Cell types can be defined by the genes expressed, and unique cell-type-specific transcriptional mechanisms control these expressions. Single nucleotide polymorphisms (SNPs) in the DNA can be associated with diseases, but approximately 95% fall in the non-coding region. Usually, it is unknown whether these variants are causal, and which gene and cell type they affect.Advances in single-cell RNA-sequencing improved our understanding of heterogeneous tissues and led to the discovery of many new cell types. This new technology also presents computational challenges including consistent cell-type annotations. It is essential to annotate cells using classification instead of currently practiced clustering methods. To facilitate this transition, we benchmarked cell-type classification methods and developed computational methods to automatically build reference atlases using multiple already labeled single-cell datasets.Next, we establish a relationship between mutations and their effect on gene or isoform expression. We study sequence-to-expression models that can predict an alteration in expression when a mutation is observed. Given that gene expression mechanisms are cell-type specific, we introduce sequence-to-expression models based on single-cell data to make cell-type-specific predictions. We use these models to show that certain mutations are indeed changing expression, increasing our understanding of transcriptional regulation. Show less
Uveal melanoma, as opposed to cutaneous melanoma, is a tumor which is predominantly located in the center of the eye. Given its potential aggressive nature with fatal consequences, it is important... Show moreUveal melanoma, as opposed to cutaneous melanoma, is a tumor which is predominantly located in the center of the eye. Given its potential aggressive nature with fatal consequences, it is important to start treatment in a timely manner. Currently, treatment consists of decreasing the size of the tumor by means of radiation therapy or surgery, but by mapping the genetic makeup of uveal melanoma, personalized treatment potentially becomes possible. In this thesis we tried to solve a piece of this genetic puzzle by focusing on genetic and cellular differences within and between different uveal melanoma. By means of an innovative new technique, digital PCR, we aimed to visualize this so-called heterogeneity. We demonstrate that we can accurately assess which genetic aberrations are present in uveal melanoma. Moreover, we show in what percentage of cells these aberrations are present. Additionally, we illustrate which cell types play an important role in uveal melanoma development. This way, we characterized a substantial amount of different uveal melanoma based on their heterogeneity profile. Lastly, we display whether and how uveal melanomas with a specific heterogeneity profile are susceptible to treatment. Show less
Cardiac arrhythmias are a common cause of sudden death worldwide. However, despite decades of thorough investigation the underlying biophysical mechanisms of cardiac arrhythmias are still... Show moreCardiac arrhythmias are a common cause of sudden death worldwide. However, despite decades of thorough investigation the underlying biophysical mechanisms of cardiac arrhythmias are still insufficiently understood due to incomplete theories and the lack of precise spatiotemporal control in experiments. In the last decade, the problem of insufficient spatiotemporal control has started to be tackled by means of a new technique, called optogenetics. This technique employs expression of light-activated proteins, which are activated or deactivated in time and space by switching on/off light (in the near-ultraviolet to near-infrared wavelength range) in specific patterns thus realizing fully biological spatiotemporal control. However, with a few notable exceptions, cardiac optogenetic studies have only confirmed previously known mechanisms and yielded no or little novel mechanistic insights. In this thesis, to fill this gap, we combined nonlinear dynamics theory, numerical simulations and optogenetic experiments with unique spatiotemporal control to theoretically predict and demonstrate novel arrhythmogenic phenomena in cardiac tissue. Thanks to the robustness of the optogenetics methods and generality of the applied theories and computations, this thesis uncovered novel mechanisms for the biophysics of cardiac tissue that are applicable to the functioning of excitable systems in general. Show less
The ideas underlying this thesis are that lipidomics may improve the diagnosis of liver metabolic diseases, and can provide further insights into the underlying pathophysiology of these diseases.... Show moreThe ideas underlying this thesis are that lipidomics may improve the diagnosis of liver metabolic diseases, and can provide further insights into the underlying pathophysiology of these diseases. However to achieve this, the measurement of lipids should be further improved by enabling comprehensive coverage, accurate identification and quantitation. Therefore, in this thesis, new analytical methodologies using LC-MS for the accurate detection and quantitation of lipids are developed. These methods are then applied to MCADD samples to identify lipid biomarkers, which are further correlated with gene expression analysis to understand the underlying biochemical pathways. Show less
This thesis studies the geometry of representation varieties and character stacks. These are spaces parametrizing the representations of a finitely generated group, typically the fundamental group... Show moreThis thesis studies the geometry of representation varieties and character stacks. These are spaces parametrizing the representations of a finitely generated group, typically the fundamental group of a compact manifold, into an algebraic group G. We provide a number of methods to study these spaces through their invariants, focussing on the invariant that takes values in the Grothendieck ring of varieties. We show how both the arithmetic method, which studies the character stacks of compact orientable surfaces through counting points over finite fields, and the geometric method, which studies these character stacks using stratifications, can be expressed in terms of topological quantum field theories. We compute explicitly the invariants in specific cases, such as for G = SL2 and for G equal to the groups of upper triangular matrices. Motivated by these applications, we develop a number of new computational tools. Finally, we study the representation varieties and character stacks of the free groups and of the free abelian groups, of finite rank. We introduce a notion of motivic representation stability for stability in the Grothendieck ring of varieties, and show that these character stacks stabilize in this sense for G equal to the general linear groups. Show less
Gynaecological carcinomas (ovarian, endometrial, cervical, vaginal, vulvar and breast) are among the ones leading to the highest cancer-related deaths worldwide. The presence of germline pathogenic... Show moreGynaecological carcinomas (ovarian, endometrial, cervical, vaginal, vulvar and breast) are among the ones leading to the highest cancer-related deaths worldwide. The presence of germline pathogenic variants in the breast cancer susceptibility genes BRCA1 and BRCA2 is associated with an increased risk for breast and ovarian cancer as well as other cancers. The BRCA proteins play a crucial role in the homologous recombination (HR) pathway, which is the only DNA damage repair pathway that can repair DNA double-strand breaks error-free. The repair of DNA damage by HR is crucial to maintain genomic integrity. Patients whose tumors are HR-Deficient (HRD) are particularly sensitive to platinum-based chemotherapy and poly-ADP ribose polymerase inhibitors (PARPi). BRCA1/2 deficiency, but also (epi)genetic defects in other HR-related genes or other unknown factors can lead to HRD. However, BRCA1/2 deficiency, and not HRD, is still the leading prerequisite to be eligible for PARPi therapy. In this thesis, two functional RAD51-based HRD tests (RECAP and RAD51-FFPE test), were developed and validated using gynaecological carcinomas. These functional HRD tests were benchmarked with BRCA1/2 deficiency and with other DNA-based HRD tests measuring genomic scars and mutational signatures. In addition, the prevalence of functional HRD was explored among a variety of gynaecological carcinomas. Show less
In this Ph.D. thesis we study the interaction of low energy electrons with thin materials, namely layered materials (graphene, hexagonal boron nitride, molybdenum disulfide) and organic films. At... Show moreIn this Ph.D. thesis we study the interaction of low energy electrons with thin materials, namely layered materials (graphene, hexagonal boron nitride, molybdenum disulfide) and organic films. At these low energies the quantum mechanical wavelength of the electron wave function is in the order of a few Angstroms, thus comparable to the interlayer distance in layered materials. This leads to resonances in the electron reflection/transmission spectrum, comparable to the interference of light when it is reflected from a thin film. We use low energy electron microscopy (LEEM) and electron Volt transmission electron microscopy (eV-TEM) to determine the energy dependent electron mean free path (MFP) and identify resonant transmission/reflection states related to the unoccupied band structure. Furthermore, we use photoemission electron microscopy (PEEM) to image low energy electrons from a gold surface covered with a film of chiral organic molecules. We image the photoexcited electrons and compare the intensity of photoemission caused by different (circular) polarizations of light. Show less
De ziekte van Parkinson treft wereldwijd miljoenen mensen. Naarmate de ziekte vordert, ervaren patiënten motorische en niet-motorische symptoomschommelingen die hun kwaliteit van leven aanzienlijk... Show moreDe ziekte van Parkinson treft wereldwijd miljoenen mensen. Naarmate de ziekte vordert, ervaren patiënten motorische en niet-motorische symptoomschommelingen die hun kwaliteit van leven aanzienlijk beïnvloedt. Ze wisselen tussen perioden van een goede respons op medicatie (ON) en perioden van onvoldoende respons (OFF). Hoewel plotselinge OFF-perioden goed behandeld kunnen worden met apomorfine subcutane injecties, is er behoefte aan gebruiksvriendelijkere toedieningsmanieren. Dit proefschrift onderzocht twee nieuwe toedieningsvormen: een inhalator met apomorfine en een spray die apmorfine-oplossing aanbrengt aan de binnenkant van de wangen. Inhalatie zorgde voor snelle opname van apomorfine in het bloed (binnen 1-2 minuten) en verbeterde Parkinson symptomen (binnen 10 minuten). Ook de apomorfine spray bereikte klinisch relevante concentraties in het bloed. Maximale concentraties werden bereikt na 32-53 minuten (verschillend per dosering). Een vervolgstudie moet nog onderzoeken hoe snel dit tot verbetering van Parkinson symptomen leidt.Om het effect van nieuwe snelwerkende medicatie te onderzoeken, zijn objectieve, kwantitatieve en snelle metingen nodig. Vooral voor de behandeling van OFF-perioden is het bepalen van aanvang van effect essentieel om te kijken of een medicijn geschikt is voor deze indicatie. Momenteel wordt vaak de MDS-UPDRS III schaal gebruikt om de effectiviteit van medicijnen te beoordelen, maar deze meting vereist een getrainde beoordelaar, duurt relatief lang (ca. 15 minuten) en vertoont variatie tussen en binnen beoordelaars. Dit proefschrift toonde aan dat wijsvingertikken en duim-wijsvingertikken korte, beoordelaar-onafhankelijke metingen zijn die het effect van (snelwerkende) dopaminerge Parkinson medicatie kunnen aantonen met een vergelijkbare effectgrootte als de MDS-UPDRS III. Machine learning identificeerde een optimale combinatie van wijsvingertik parameters die zelfs beter is in het aantonen van medicatie effecten dan de MDS-UPDRS III. Show less
Spectroscopic studies on fluorescent single molecules in organic condensed matter does not only provide information about the molecule itself, but also its near environment. By suppression of... Show moreSpectroscopic studies on fluorescent single molecules in organic condensed matter does not only provide information about the molecule itself, but also its near environment. By suppression of phonon-induced broadening of spectral lines through cooling to low temperatures, small changes in the spectral lines’ position can be observed in response to weak variations in local fields. These variations can for instance be caused by rearrangements of charges or minute changes in the crystal lattice around the molecule. Therefore, molecules are sensitive sensors to what happens at the nanoscale. This is exemplified by coupling to an external electric field, inducing a Stark shift of the molecule’s spectral lines, as shown in Chapter 4. Other dynamics, related to the crystal around the molecule, are resolved in the fluorescence of molecules on the surface of two-dimensional hexagonal boron nitride, shown in Chapter 5. In Chapter 2, 3 and 6, perylene molecules are studied in a new host crystal with the aim of detecting a ‘forbidden’ transition to the triplet state from the ground state, a transition required for building a single-molecule optical switch. Show less
In this thesis we study curves. In the first half, we study moduli spaces of curves and Gromov-Witten invariants, certain kinds of curves counts. We employ logarithmic geometry for this. Some major... Show moreIn this thesis we study curves. In the first half, we study moduli spaces of curves and Gromov-Witten invariants, certain kinds of curves counts. We employ logarithmic geometry for this. Some major results include the polynomiality of the double ramification cycle and recursive relations for the log double ramification cycle.In the second half we study rational points on curves, in particular Chabauty's method for finding the rational points and extensions of it. Major results include that the geometric (quadratic) Chabauty method is theoretically stronger than the original (quadratic) Chabauty method, and that local heights for quadratic Chabauty are explicitly computable. Show less
