High-level software testing is often complex and highly situational, and there are areas where conventional methods of automation are simply not feasible. One such potential area is the post-mortem analysis of failed tests. This thesis explores the feasibility of using Large Language Models (LLMs) for automating the analysis of simulation-based test-scenarios within Tacsi, a tactical simulation environment developed by Saab AB. A system monickered ScenarioLens was implemented to summarise JSON-encoded recordings of these scenarios and to generate a conclusive briefing. Extensive experiments evaluated model size and prompting strategies, validated through a novel method of measuring accuracy tailored to the LLM-generated conclusions. Results demonstrate that LLMs are able to generalise well to this domain under certain constraints and offer promising avenues for integrating AI into high-level, domain-specific software testing workflows. Additionally, the thesis explores the use of multimodal LLMs for the same task, though this approach showed less potential compared to purely text-based methods.

Multi-modal retrieval has traditionally focused on combining diverse query in-puts, such as text and sketches, in remote sensing and computer vision. However, retrieval involving multi-modal target representations, such as paired rgband depth data, remains largely unaddressed. This work investigates whetherincorporating a depth modality can improve the performance of vision-languagemodels in the context of satellite image retrieval. To explore this, a novel dataset, rsitdd, was constructed, combining orthophotos and digital height models, andused to train a clip-based remote sensing depth encoder. Experimental results show that models augmented with a depth encoder outperform their text-image-only counterparts across multiple benchmark settings. These findings highlight the potential of depth-enhanced models for remote sensing applications and demonstrate that even simple fusion techniques can yield measurable performance improvements.

Tantalum nitride (Ta3N5) is a promising semiconductor for solar-driven photoelectrochemical (PEC) water splitting, but its performance is limited by intrinsic defects. Here, we investigate the effect of titanium (Ti) doping (0–10 at%) on the structural, compositional, and optoelectronic properties of Ta3N5 thin films. At low concentrations (<2 at%), Ti4+ preferentially substitutes Ta at four-coordinated sites, enhancing nitrogen incorporation and suppressing defect states associated with under-coordinated Ta. This leads to improved carrier dynamics and prolonged electron–hole lifetimes. Higher doping levels (≥3.5 at%) result in occupation of three-coordinated sites, inducing increase in the oxygen content, lattice distortion, and defect formation that deteriorate carrier lifetimes. PEC measurements reveal that optimized Ti doping significantly reduces charge transfer resistance and nearly seven-fold increase in the photocurrent. These findings underscore the importance of controlled Ti doping for defect engineering and band structure tuning to boost the PEC performance of Ta3N5 thin films.

Electrification is increasingly being adopted in mobile machinery as a means to reduce carbon emissions and improve energy efficiency. While electric solutions for actuation are growing, hydraulic systems still offer a favourable balance between cost, power density, reliability, and overall performance. Their primary drawback is efficiency, as they often rely on throttling control, which incurs high losses. However, electrified vehicles bring new opportunities for redesigning hydraulic systems.

An electric vehicle does not require a centralised hydraulic system. Integrated electric machine and hydraulic pump/motor units have long been available, enabling systems composed of multiple smaller decentralised components. Such arrangements allow the system to match flow and pressure demand directly, reducing or eliminating throttling losses.

This enables new hydraulic architectures that require alternative control methods and have not been previously analysed. This thesis focuses on the Multi-Pump System (MPS), which uses multiple smaller fixed-displacement hydraulic machines to supply flow to the actuators through on/off valves. The design minimises throttling by using the valves for flow routing and the pump/motors for control. Any pump/motor port can connect to any actuator chamber, and the architecture enables energy regeneration and recuperation.

This system offers considerable flexibility in performing a task. For instance, it can control actuators independently using varying numbers of active pump/motors, short-circuit actuator chambers to reduce the total required pump flow, or utilise the return flow for electric energy recuperation. This licentiate thesis investigates the decision-making process involved in selecting among these operating modes.

It proposes an optimisation-based method to infer viable and preferred control decisions from the actuators’ operating points, thereby reducing the control decision space. A structurally simple system is analysed, and a visualisation method is introduced to summarise the transition regions between operating modes for the hydraulic machines. Finally, a dynamic model is tested using the decisions from this analysis to develop the control system. The results indicate that this approach can be extended to more complex systems, although new strategies may be required to identify mode transition patterns.

Democracy, a contested ideal, has seen its meaning expand and contract in response to historical events, institutional change, and scholarly priorities. Despite extensive theoretical work on democratic models, a significant challenge remains in systematically quantifying how abstract political concepts evolve semantically over time, as traditional approaches have relied primarily on qualitative case studies or theoretical frameworks without empirical measurement of linguistic patterns. Yet little is known about how the term itself has evolved within political science discourse in the recent decades. This research is particularly significant given the global phenomena of democratic erosion and the increasingly multidimensional nature of democratic theory that has developed in scholarly discourse over recent decades. This thesis applies a relational semantic framework, treating concepts as dynamic networks of associations, to over 6,000 articles published between 1971 and 2024 in the American Journal of Political Science (AJPS) and the British Journal of Political Science (BJPS). Using diachronic word-embedding models trained on bootstrapped sub-corpora, five semantic dimensions are constructed, electoral, liberal, participatory, deliberative, and egalitarian, based on anchor words drawn from the Varieties of Democracy framework and Democratic Erosion Event Dataset codebook. I then measure how closely “democracy” and a broader democracy–authoritariandimension align with each dimension over time and across journals. The findings reveal that while electoral principles remain foundational to democracy’s scholarly conceptualization, a significant expansion has been incorporating liberal, participatory, and egalitarian dimensions. Academic discourse mirrors real-world democratic decline through the weakening association with deliberation and increased usage of terms related to erosion and backsliding. Journal-specific differences highlight distinctive normative priorities in how democracy is framed against authoritarianism. These semantic shifts illuminate how scholarly understandings of democracy evolve in response to both intellectual developments and global political challenges, demonstrating the value of computational approaches to conceptual analysis in political theory.

Background

This thesis investigates the multifaceted experiences and outcomes associated with childhood functional constipation, a condition that significantly affects children’s health and family life. Drawing on one quantitative and three qualitative studies, the research provides an increased understanding of how children and families navigate the challenges of diagnosis, treatment, and daily life with a child suffering from functional constipation.

Aim

The overall aim was to inform a more child- and family-centred perspective to functional constipation care by exploring the experiences of parents and children, and to evaluate a clinical treatment programme.

 Methods

Study I is a retrospective cohort review of medical records with pre-post intervention measurements. Children participating in a structured bowel management programme in a secondary outpatient clinic between 2015 and 2022 were recruited. Clinical variables were compared between the recovered and the unrecovered groups to identify predisposing characteristics and predictive factors for recovery.

Studies II and III have a qualitative design, presents interviews with parents about their experiences of having a child with constipation and of giving treatment at home. The same data set of 15 parents were used for both studies. The Reflective Lifeworld Research approach originating from phenomenology was used during analysis.

Study IV, also qualitative, presents interviews with 20 children aged between 6 and 14 about their experiences of having constipation and receiving treatment. Reflexive thematic analysis was used during analysis.

Findings

The first study demonstrates significant improvements in bowel function and symptom relief among children following the intervention. Structured care helped 44% of therapy-resistant children to reach cure or self-management abilities within six months. The study underscores the value of integrating clinical protocols with family education and support.

The second study reveals how childhood constipation disrupts family routines, emotional well-being, and social interactions as treatment support requires considerable attention and strength. Parents report feelings of frustration, helplessness, and isolation, often compounded by limited support from healthcare professionals. The findings highlight the need for more empathetic child and family-centred care strategies.

The third study uncovers the emotional and practical challenges involved in managing medication routines, feelings of abuse and inadequacy, while trying to maintain adherence. The study emphasises the importance of clear communication and sustained support to empower parents in their caregiving roles because they might question their parental identity.

The fourth study presents the child’s perspectives on functional constipation with associated treatments. It presents a close and realistic narrative of procrastination of toilet visits, fear of exposure of leakage in social situations, rectal enemas as both awful and relieving, and hope, while striving for control.

Conclusions

Together these studies offer a holistic understanding of childhood functional constipation, bridging the gap between clinical outcomes and family experiences. Families deal with physical symptoms, psychological defences, fear of social judgment, and struggles with treatment. This thesis offers actionable insights for children, parents, clinicians, and researchers to improve the quality of constipation care for children and their caregivers.

Sport is more than just a game—it's a global phenomenon that shapes cultures, economies, and communities. Football, the world's most popular sport, is a prime example. Yet beneath the surface lies an overlooked environmental cost. As the climate crisis accelerates, the sprawling network of football facilities—stadiums, training grounds, and infrastructure—emerges as a silent contributor to environmental degradation and the transgression of planetary boundaries. Two common types of fields exist: artificial and natural turf. Research on environmental impacts of these turfs remains limited, especially in cold climates. This study presents a life cycle assessment of 1 m2 artificial and natural football turfs in Nordic climates, evaluating their environmental impacts such as global warming potential, eutrophication potential and ecotoxicity potential across construction, use, maintenance, and end-of-life phases over operational lifespans of 10, 20 and 30 years. Natural turf exhibited the highest overall environmental impacts over the operational lifespan, e.g. the global warming potential was 30.6 kg CO2 eq/m2 while the artificial turf reached 15.6 kg CO2 eq/m2. During the construction phase, artificial turf generated significant emissions, mainly from material production. In the use phase, natural turf showed the greatest impacts due to diesel consumption and fertilizer application. At the end-of-life stage, artificial turf's sand and infill were reused, while the turf carpet and shock pad were incinerated for energy recovery. However, without recycling, artificial turf would represent the highest environmental burden among the evaluated alternatives. Implementing effective recycling and energy recovery strategies is essential to mitigate its environmental impact. Furthermore, sourcing turf materials locally, combined with substituting conventional maintenance equipment with electric robotic alternatives, can further reduce overall environmental impacts.

The sustainable governance of ocean spaces remains a pressing challenge for Small Island Developing States (SIDS), where fragmented and sector-based management has long constrained the effectiveness of ocean governance. In these settings, multiple institutions often operate in isolation, pursuing competing priorities with limited coordination, resources, and accountability. These issues are particularly acute in the Pacific Islands, where vast ocean territories, customary marine tenure, and deeply rooted cultural values intersect with complex governance realities shaped by political and market-oriented forces. In response to these challenges, Marine Spatial Planning (MSP) offers an ecosystem-based, integrative framework for reconciling ecological, social, and economic objectives in marine and coastal governance. Yet, despite its growing adoption, MSP in SIDS remains a relatively new and complex endeavour, reflecting diverse institutional and cultural realities. This compilation thesis critically examines the role of MSP in advancing ocean governance in Pacific SIDS, with Fiji as the focal empirical case. It addresses a key knowledge gap regarding the contextual conditions under which MSP can function effectively as a governance instrument in resource-constrained, socially diverse settings. The four papers collectively trace this inquiry. The first paper identifies enabling and constraining conditions for MSP in SIDS through a semi-systematic literature review; the second explores how MSP could respond to Fiji’s ocean governance challenges; the third investigates how Fiji’s MSP process aligns with Indigenous coastal perspectives; and the fourth evaluates the extent to which MSP is fit for purpose in advancing transformative ocean governance across the Pacific.

The findings demonstrate that MSP’s effectiveness in SIDS depends on robust collaborative mechanisms underpinned by facilitative leadership, clearly defined legal mandates, strong institutions, and social capital that bridges formal and informal governance systems. Yet, the Pacific context presents distinct governance challenges, including colonial legacies of centralized authority, limited stakeholder ownership, overlapping institutional responsibilities, and low public awareness of MSP that constrain the transformative potential of MSP and risk perpetuating top-down approaches. Even so, MSP in this region has the potential to become a transformative governance instrument, precisely because of the Pacific’s remarkable ecological, cultural, and institutional diversity. Rather than a barrier, this diversity could serve as the foundation for a collaborative, polycentric, and adaptive governance architecture, in which multiple islands, customary authorities, and local institutions operate as interconnected nodes of innovation, learning, and resilience. In practical terms, this transformation requires legal recognition of customary marine tenure, resourcing local communities for meaningful participation, and empowering boundary organizations and communities of practice to bridge governance processes across scales. Together, these measures outline a pathway for MSP in the Pacific to move beyond its current technocratic and procedural emphasis, evolving into a relational, inclusive, and adaptive governance process capable of aligning ecological sustainability and economic aspirations with cultural legitimacy, equity, and social justice.

Marine Spatial Planning (MSP) is a globally established tool to support integrated ocean management. As Small Island Developing States (SIDS) embrace MSP, this study focuses on Fiji as it begins its MSP process alongside the implementation of newly established ocean-related policies and legislation. The study investigates whether MSP has the potential to address the challenges identified by various actors and, if so, to explore how this can be achieved. A mixed-methods case study approach was employed, combining analysis of national policy documents with semi-structured interviews involving key actors in Fiji’s ocean governance. Guided by collaborative governance theory, interview data were thematically analyzed using NVivo. The results indicate that MSP could serve as a viable governance tool capable of addressing institutional silos and overcoming coordination challenges. However, prevailing interagency conflicts, power imbalances between stakeholders, and ambiguity surrounding roles and responsibilities pose significant barriers to meaningful participation, risking the reinforcement of the status quo. Our findings highlight that effective leadership, inclusive governance arrangements, financial sustainability, and social capital are mutually reinforcing enablers of collaboration within MSP. Finally, we recommend a staged, evidence-based approach to institutional reform aligned with fiscal and political realities that offers the most viable pathway to building a legitimate, resilient, and durable MSP system

This work defines topological properties and aspects of merge trees which are relevant for detection of topological symmetry in merge tree representations of crystalline materials. For this, three distinct perspectives on symmetry axes have been defined which are relevant in differentiating symmetry in different contexts. Namely, explicit axis, semi-implicit axis and implicit axis. It further motivates why symmetry detection approaches based on explicit axis perspective are not applicable to the context of topological structures in crystals due to the rearranging of initially symmetrical hierarchies. Instead, an alternative method based on implicit axis perspective is suggested for detecting symmetries in merge tree representations of crystalline materials. It is also evaluated why merge tree representations only support partial detection of specific larger structures such as chains and layers in crystals.

Computer vision aims to equip machines with perceptual understanding—detecting, recognizing, localizing, and relating visual entities to existing sources of knowledge. Machine learning provides the mechanism: models learn representations and decision rules from data and are expected to generalize beyond the training distribution. These systems already support biodiversity monitoring, autonomous driving, and geospatial mapping. In practice, however, textbook assumptions break down: the concept space is vast, data is sparse and imbalanced, many categories are rare, and high-quality annotations are costly. In addition, deployment conditions shift over time—class frequencies and visual domains evolve—biasing models toward frequent scenarios and eroding reliability.

In this work, we develop methods for training reliable visual recognition models under more realistic conditions: class imbalance, limited labeled data, and distribution shift. Our contributions span three themes: (1) debiasing strategies for imbalanced classification that remain reliable under changes in class priors; (2) semi-supervised learning techniques tailored to imbalanced data to reduce annotation cost while preserving minority-class performance; and (3) a unified multimodal retrieval approach for remote sensing (RS) that narrows the domain gap.

In Paper A, we study long-tailed image recognition, where skewed training data biases classifiers toward frequent classes. During deployment, changes in class priors can further amplify this bias. We propose an ensemble of skill-diverse experts, each trained under a distinct target prior, and aggregate their predictions to balance head and tail performance. We theoretically show that the ensemble’s prior bias equals the mean expert bias and that choosing complementary target priors cancels it, yielding an unbiased predictor that minimizes balanced error. With calibrated experts—achieved in practice via Mixup—the ensemble attains state-of-the-art accuracy and remains reliable under label shift.

In Paper B, we investigate long-tailed recognition in the semi-supervised setting, where a small, imbalanced labeled set is paired with a large unlabeled pool. Semi-supervised learning leverages unlabeled data to reduce annotation costs, typically through pseudo-labeling, but the unlabeled class distribution is often unknown and skewed. Naïve pseudo-labeling propagates the labeled bias, reinforcing head classes and overlooking rare ones. We propose a flexible distribution-alignment framework that estimates the unlabeled class mix online and reweights pseudo-labels accordingly, guiding the model first toward the unlabeled distribution to stabilize training and then toward a balanced classifier for fair inference. The proposed approach leverages unlabeled data more effectively, improving accuracy, calibration, and robustness to unknown unlabeled priors.

In Paper C, we move beyond recognition to unified multimodal retrieval for remote sensing—a domain with scarce image–text annotations and a challenging shift from natural images. Prior solutions are fragmented: RS dual encoders lack interleaved input support; universal embedders miss spatial metadata and degrade under domain shift; and RS generative assistants reason over regions but lack scalable retrieval. To overcome these limitations, we introduce VLM2GeoVec, a single-encoder, instruction-following embedder that aligns images, text, regions, and geocoordinates in a shared space. For comprehensive evaluation, we also propose RSMEB, a unified retrieval benchmark that spans conventional tasks (e.g., classification, cross-modal retrieval) and novel interleaved tasks (e.g., visual grounding, spatial localization, semantic geo-localization). In RSMEB, VLM2GeoVec narrows the domain gap relative to universal embedders and matches specialized baselines in conventional tasks in zero-shot settings. It further enables interleaved spatially-aware search, delivering several-fold gains in metadata-aware RS applications.

“Cook and eat” meal kits from pre-packaged food delivery services provide exact portions of ingredients required for specific recipes, promoting sustainability by eliminating food waste in the cooking process. However, these services often overlook some of the current consumer behaviours associated with their use. This study aims to explore consumer behaviour regarding the use of meal kits in the Swedish context, with a particular focus on the sustainability of these behaviours. Applying activity theory as the theoretical framework, this study explores how consumers interact with meal kits during the ordering, cooking, storing and disposal and identifies potential areas for service and packaging improvement to promote sustainability. The methodology used includes a combination of qualitative survey (24 answers), in-depth interviews and observations with 4 households to gather data about consumer cooking practices and perceptions and self-reporting from 3 of the authors. Specifically, the study focuses on the operational layer of activity theory, analysing the routine actions and adjustments consumers make while interacting with meal kits, which directly and both consciously and unconsciously impact food waste and its sustainability. Considering the harmonies and tensions found between the consumer, the meal kit, including the recipe, the food, and the packaging, and the homemade meal, the findings show that while meal kits ease the cooking process and save time, they are perceived as having excessive packaging and do not completely eliminate food waste as promoted. Design suggestions include considering returnable packaging systems, offering ingredient replacement options, providing freezable and microwavable recipes, and including new cooking guidelines such as taste-testing steps. These changes can better align them with consumer needs, and potentially reduce the impact of consumer use, reducing food waste and increasing the sustainability perception of the system.

Modern clinical decision support requires models that are both accurate and mechanistically interpretable. DNA methylation tracks the cumulative influence of development, lifestyle, and environment on gene regulation, but its dimensionality and tissue specificity complicate analysis and clinical application. This thesis develops explainable deep learning methods that learn coherent biological signals from genome-wide methylation data, aiming to derive reliable biomarkers of aging, disease risk and severity, and system-level health. Central to our approach are deep autoencoders, unsupervised multi-layered neural networks that efficiently compress DNA methylation data into low-dimensional embeddings that preserve relevant biology, paired with interpretability techniques that expose feature contributions and model reasoning, such as perturbation-based latent activation.

By training on large multi-tissue compendia of human DNA methylation samples, we observed that the autoencoders self-organized their latent spaces, recapitulating protein-protein interaction (PPI) modules. Interpreting these structured embeddings yielded pathway-enriched epigenomic signatures that supported accurate epigenetic age estimation and robust classification of disease status and smoking. Building on these findings, we introduced a PPI-guided autoencoder that incorporates a graph-regularized protein interaction prior, encouraging each latent unit to be functionally specific and colocalized within the human interactome. We showed that this soft guidance improved the mechanistic interpretability of downstream models, in this case supervised translators that map between omics modalities (transcriptomics, DNA methylation, genomics).

In parallel, we combined autoencoder embeddings with established aging markers to train explainable neural-network age clocks that achieved state-of-the-art cross-tissue precision, while also capturing fine-grained developmental, immune, and metabolic signatures. Finally, we operationalized these representations in a clinical decision-support pipeline that predicts respiratory, cardiovascular, and metabolic system-level health scores from blood methylation, with supervised deep learning models that highlight biological processes associated with each physiological system. Collectively, this work provides a scalable and auditable framework that converts methylomes into interpretable feature sets and actionable indicators for clinical use, enabling early risk assessment, monitoring of treatment responses and lifestyle changes, and informed therapeutic target prioritization.

Three approaches for Doppler radar are compared with respect to anti-jamming: A non-adaptive one which maintains a constant transmission frequency, an simple adaptive one that changes frequency based on current interference on the spectrum, and a reinforcement-learning based adaptive one which utilises Q-learning to pick the most appropriate frequency band based on the current state of the spectrum. Both simulated and real-world tests with a doppler radar are carried out to assess which approach is more successful, and to determine appropriate values for the learning rate and discount factor in the reinforcement learning-based approach. Tests show that the reinforcement learning approach is more successful both in not being jammed, and (for the adaptive solutions) in transmitting on the same frequency for longer, which enables detection of slow targets.

Syftet med projekt Meningsskapande i Naturvetenskap och Teknik genom Skönlitteratur (MNTS) var att utveckla undervisningen i naturvetenskap och teknik i förskolan och åk F-6 genom att integrera skönlitterära böcker i undervisningen. Projektet var ett forsknings- och utvecklingsprojekt där samverkan var en viktig del. Samverkan byggde på ett samarbete mellan forskare vid Linköpings universitet, medarbetare från Utbildningskontoret, Norrköpings stadsbiblioteks medarbetare, skolbibliotekarier samt förskollärare och grundskolelärare i ett antal av Norrköpings kommunala skolor och förskolor. Projektet har finansierats av Norrköpings fond för forskning och utveckling.

Fault diagnosis is essential for ensuring the safety and reliability of complex engineering systems, where faults in individual components can propagate into degradation of overall process performance and eventually failure. Traditional model-based diagnosis relies on mathematical models to generate residuals that capture deviations from expected behavior, while data-driven diagnosis uses machine learning on historical data to detect or classify faults. While both approaches aim for fault detection and isolation, achieving satisfactory performance is a non-trivial task. Model-based methods require detailed domain expertise and accurate system models, whereas data-driven approaches often struggle with scarce or unrepresentative fault data and unreliable generalization beyond training distributions.

A primary focus of this thesis is the investigation of hybrid fault diagnosis methods that integrate physical insight with deep learning architectures to reduce reliance on fault data. To address this, structural analysis is employed as a foundational tool for identifying redundancy, and computational sequences derived from the structural model are used to guide the design of neural network-based residual architectures. Additionally, the residuals are designed to respect the underlying differential relationships of dynamic systems, with careful consideration given to the numerical evolution of system states. Several sequential modeling approaches are implemented, and a methodology for generation, training and assessment of these hybrid diagnostic systems is provided.

Another aspect of this thesis is addressing model inaccuracies in data-driven diagnosis. Diagnosis systems must handle noisy sensor signals and incomplete training data, which can cause unreliable or overconfident diagnostic statements. To address this, the thesis explores different techniques to evaluate the reliability of the data-driven residuals. One approach is to model validity regions of residual models in which diagnostic conclusions remain reliable. The second approach uses probabilistic ensemble neural networks to quantify aleatoric and epistemic uncertainty and to handle model inaccuracies. In this thesis, it is shown how to integrate data-driven models and the assessed reliability measures, into a consistency-based diagnosis framework.

This thesis contributes to bridging the areas of model-based and data-driven fault diagnosis. The proposed methods have been validated using data from both simulations and real automotive case studies. The results show that the design of neural network-based residuals can reason about abnormal behavior in complex dynamic processes even when there is limited or no training data from faults.

This study examines trans-boundary air pollution by a coal power plant near the India-Bangladesh border. Given the well documented negative health effects by air pollution, especially particulate matter, it is imperative that this issue is thoroughly investigated. This study adds to an existing body of knowledge around the FSTCPP and aims to further expand the understanding of the emissions on human health.  

 This study used three methods to accomplish the goal. The first method used acellular assays. 1) a dithiothreitol assay to measure oxidative potential, and 2) a hydroxyl assay to measure hydroxyl radical production. These two assays tested particulate matter on samples collected in Bangladesh. The second method involved correlation analyses of weather variables (wind direction, and temperature) and lung health data. The third method involved a cellular assay to measure the toxicity of selected metals to cells.  

 The results from the assays showed that even after accounting for inter-lab variability, the Bangladeshi sites had higher oxidative potential (OP) and hydroxyl (OH) generation capacity than their Indian counterparts. Temperature showed an inverse correlation with OP/OH for the most part, with wind direction having a positive correlation with volume-based OP/OH, and less correlation with intrinsic OP/OH. No correlation was found between OP/OH values and spirometry data. Finally, the cellular exposure assay showed that for certain metals, e.g. Se was more toxic, and cell count decreased as concentration increased.  

This thesis investigates student perspectives on climate adaptation. As climate change progresses, rising temperatures are expected, something that requires both mitigation and adaptation. This case study invited students at Linköping University, who are frequent users of the campus environment, to participate in a photovoice study to identify strengths and problems in the campus's capacity to handle rising temperatures. Furthermore, the analysis explores how students perceive climate adaptation through the lens of framing theory and discusses their experiences with the photovoice method. The results reveal that a range of strengths and weaknesses were identified within the campus environment. However, no single adaptation measure was perceived as entirely positive or negative. Instead, effective climate resilience demands active management and a combination of various adaptation measures. These measures manifested both in physical forms, such as nature-based solutions and material choices, and in non-physical measures, such as the dissemination of information. Additionally, the findings indicate that students employed three distinct frames in their understanding of climate adaptation: climate adaptation as a multifunctional environment, climate adaptation as a product of environmental cues, and climate adaptation as a nudge for behavior change. The methodology provided students with the opportunity to share their narratives, encouraged critical reflection on their surroundings, and fostered personal growth. The thesis concludes with a discussion that situates the results within a broader context, positioning this case in relation to its character as an example of climate adaptation in an educational environment.

Lead-free halide double perovskites (elpasolites) have emerged as promising alternatives to lead halide perovskites due to the toxicity of lead for various optoelectronic applications, including photodetectors, light-emitting diodes, photocatalysts, and spintronic devices.

This research focuses on fully inorganic chloro-complex compounds with the general formula Cs₂Ag_xNa_1-xInCl₆ (x=0…1) as host materials doped with transition metal ions V³⁺ or Cr³⁺. These systems are of particular interest due to their broadband near-infrared luminescence properties and potential applications in optoelectronic devices. Additionally, this work investigates isostructural transition metal-based double perovskite compounds Cs₂(Ag/Na)FeCl₆ that exhibit antiferromagnetic ordering, making them promising candidates for spintronics and quantum information technology. To probe the atomistic structure of these materials, this thesis employs multiple advanced characterization techniques. Synchrotron- based X-ray photoelectron spectroscopy (XPS), near-edge X-ray absorption fine structure (NEXAFS), extended X-ray absorption fine structure (EXAFS), solid-state nuclear magnetic resonance (ssNMR), electron paramagnetic resonance (EPR), and optical spectroscopy were used to examine the dispersion, oxidation state, and local symmetry of dopants and host constituents at the nanoscale. The findings reveal that transition metal ion incorporation may deviate from ideal substitution, presenting challenges for certain applications while simultaneously enabling novel functionalities.

Overall, this work contributes to the fundamental understanding of structure-property relationships in lead-free halide double perovskites and establish a foundation for the rational design of next-generation optoelectronic and spintronic materials with enhanced performance and reduced environmental impact.

Introduction A healthy diet improves glycaemic control and reduces cardiovascular risk in type 2 diabetes (T2D). However, access to dietitians is limited. Several countries have implemented mandatory interpretive front-of-pack labelling to guide consumers towards healthier food choices, but Sweden has not. Smartphone applications may offer an alternative platform to provide such information. This study evaluates the dietary and clinical impact of a novel application providing interpretive labelling to Swedish adults with T2D.

Methods and analysis This is a fully decentralised randomised controlled trial. 900 individuals with T2D for ≥2 years who regularly shop for groceries will be recruited via general practices and community advertisements. Participants will be randomised to receive either: (1) access to the FoodSwitch mobile application plus standard written dietary advice, or (2) standard written dietary advice only. The FoodSwitch application allows users to scan barcodes on packaged foods to receive recommendations of healthier alternatives within the same category. The primary outcome is the difference in change in mean self-measured glycated haemoglobin between groups after 6 months. Secondary outcomes include differences in changes in waist circumference, body weight, quality of life, medication use, hospitalisations and all-cause mortality at 26 weeks. Exploratory outcomes include omics analyses. Recruitment is ongoing. Expected study completion on 31 December 2026.

Ethics and dissemination The trial has received ethical approval from the Swedish Ethical Review Authority (2023-06622-01, 2024-06668-02, 2024-07357-02 and 2025-01095-02) and is performed in line with World Medical Association Declaration of Helsinki and the General Data Protection Regulation. Results will be published in a peer-reviewed international journal.

Social-influence processes not only affect the rate at which behaviors spread but can also decouple adoption behavior from individual preferences, and thereby bring about unexpected collective outcomes that cannot be predicted on the basis of the initial likes and dislikes of the individuals involved. However, the conditions under which social influence can lead to such decoupling are not well understood. We identify a social-influence mechanism that widens individuals’ behavioral repertoires and breaks the link between individuals’ initial preferences and the collective outcomes they jointly bring about. We test the micro-level assumptions of the mechanism in the context of cultural choices on Spotify, combining topic modeling with traditional statistical matching to cultural change. agent-based simulation estimate peer-to-peer influence effects from digital trace data. We then use agent-based simulations to examine the macro-level consequences of “wide” social influence and its importance for explaining cultural change.

This thesis explores how traditional upholstery techniques can reveal the hidden qualities of the craft when applied in unexpected contexts. Despite the upholsterer's central role in furniture history and the comfort of everyday life, the craft is often overlooked and misunderstood, as the majority of the work is concealed beneath the fabric.

Through practical work and historical research, the project has explored how upholstery techniques can be used beyond traditional seating furniture. The main outcome of the work is a functional console table that demonstrates traditional techniques such as roll edge padding, blind stitching, piped edge, leather binding and braided webbing in an innovative way. The table is inspired by baroque triad furniture ensembles.

The study is based on qualitative research methods with a focus on practice-based research, documented through photographs and a logbook. Literature studies of historical and contemporary sources have provided theoretical foundation and historical context. William Morris and Jean Baudrillard's thoughts and perspectives on craft and mass production have contributed to shaping the work's theoretical framework.

The results show that the traditional invisibility of the upholstery craft can be turned into an opportunity for innovation. The project demonstrates how making craft techniques visible can contribute to increased understanding and appreciation of the profession. The work opens up new possibilities for developing the upholstery craft, while preserving and highlighting traditional knowledge for future generations.

Amphibians have a three-chambered heart, and the mechanics of living with only one ventricle and effectively separating deoxygenated blood and oxygenated blood have been difficult to understand. The outflow tract conus arteriosus contains the structure septum coni, that directs deoxygenated blood to the pulmocutaneous arch and oxygenated blood to the aortic arch, minimizing mixing. In 2021, high-resolution X-ray micro-CT imaging provided new insights into the internal structure of the conus arteriosus, revealing an unexpected connection between the right atrioventricular valve and the conus arteriosus. This connection allows blood separation to occur without a fully divided ventricle, providing a clearer understanding of amphibian cardiac physiology. This review aims to explore these findings and the development of the conus arteriosus from tadpole to adult frog.

This compilation thesis problematises challenges of inclusion in the Swedish welfare state, which delivers welfare services through digital government. This is done by analysing individuals’ practices of social rights and responsibilities among residents in multi-ethnic and socioeconomically weaker neighbourhoods. The analysis pays particular attention to people’s individual capabilities that enable such practices. The study builds on four papers.

The utilisation of digital government comes with ambitions of inclusion and efficiency to strengthen democracy, support citizens’ rights and participation, and provide access to public services. However, this is not always the case. Instead, through digital divides, exclusion is increasing and altering how people use welfare services, interact with the government, and exercise their citizenship rights and responsibilities. Digital inequalities pose a barrier for the welfare state to uphold its ambitions. This thesis develops a theoretical framework grounded in theories of citizenship and digital citizenship, incorporating digital inclusion, administrative literacy, and language proficiency.

This thesis focuses on how increasing socioeconomic inequalities and digitalisation are expressed among inhabitants in segregated neighbourhoods with lower income, education, and a higher share of migrants with refugee backgrounds. It examines this in the context of public provision delivered through digital services. Analysing the experiences of residents in vulnerable neighbourhoods presents methodological challenges of representation and measurement, including linguistic barriers, social desirability bias, recruitment challenges, and issues of trust. Therefore, this thesis draws on an immersive sequential mixed-methods field study conducted in two diverse Swedish neighbourhoods. It includes field visits, a survey, observations, focus groups, and interviews with both residents and professionals working in or with these deprived neighbourhoods.

By studying a hard-to-survey population, residents in disadvantaged neighbourhoods, the thesis not only contributes to the lack of data and methods for gathering this data but also problematises theoretical understandings of (digital) citizenship and inclusion. It shows that citizenship inclusion in a (digital) welfare state depends on the interplay of social rights, responsibilities and practices. While formal rights and responsibilities enable access and legal inclusion, meaningful inclusion requires the practice of them, which depends on people’s individual capabilities and practised inclusion. The thesis highlights residents’ challenges in practising social rights and responsibilities through the use of public digital services, mainly due to a lack of digital skills, administrative literacy, and language proficiency.

The purpose of this thesis is to develop and evaluate an AI-driven work preparation generator to improve efficiency and quality in producing work preparations in the construction industry. The research was conducted in collaboration with Stångebro Bygg AB and involved a combination of qualitative and quantitative methods. Two prototype solutions were developed using generative AI – one based on OpenAI’s language model (GPT) and another using Microsoft Azure’s AI services – and their outputs were tested on real project scenarios. These AI-generated work preparation documents were then compared to traditional, manually prepared documents in terms of content quality, efficiency, and usability. Additionally, interviews with industry practitioners (both within the company and externally) were carried out to gather insights on current work preparation practices and expectations from AI tools. The results demonstrate that both AI prototypes successfully generated work preparations closely matching the structure and content of those produced manually. The OpenAI GPT-based solution provided more comprehensive initial drafts, while the Azure-based solution delivered outputs more directly focused on the user’s specific inputs. In comparison to the manual process, the AI-driven approach significantly reduced the time required to produce a work preparation and increased the level of standardization in the documentation. However, some human review was still necessary to adjust the AI outputs to project-specific conditions. The study identified key requirements for an effective AI-based tool: alignment with the company’s templates and terminology, a user-friendly interface adaptable to practitioners’ needs, and integration of domain-specific knowledge of local construction regulations. In conclusion, the thesis finds that generative AI holds strong potential to streamline and enhance the work preparation process in construction. The optimal outcome is achieved by combining AI capabilities with human expertise – AI contributes speed and structured content, while human professionals ensure context relevance and quality control. This work thereby confirms that the developed AI-driven generator can be a functional, useful, and integrable solution, leading to more efficient preparation workflows and improved document quality in construction projects.

Background: Emotion regulation involves managing which emotions we experience, when and how we feel them, and how we express them. Difficulties in emotion regulation, known as emotion dysregulation, are central to many mental health conditions. With significant overlap across psychiatric diagnoses, transdiagnostic approaches to understanding and treating these issues are gaining attention. Adolescence is marked by major physical, psychological, and social changes, alongside heightened emotional experiences and the onset of many psychiatric disorders. This highlights the importance of addressing emotion dysregulation in child and adolescent psychiatric care, and the need to evaluate its assessment and explore transdiagnostic interventions.

The overarching aim of this thesis was twofold: first, to examine the assessment of emotion dysregulation in adolescents within child and adolescent psychiatric services; and second, to evaluate the feasibility, outcomes and experiences of a brief adjunctive emotion regulation skills training.

In Study I, the psychometric properties of a brief version of the Difficulties in Emotion Regulation Scale (DERS-16; Bjureberg et al., 2016) were examined in a child and adolescent psychiatric sample (n = 281). The DERS-16 was also used to assess emotion dysregulation in a community sample (n = 3,169). A two-step cluster analysis was conducted to identify subgroups within the clinical sample. Study II was a pilot study evaluating a brief emotion regulation skills training group delivered jointly to adolescents (n = 20) and their parents (n = 21) in an outpatient child and adolescent psychiatric setting. The aim was to assess the feasibility and outcomes of the intervention using a within-group design. Pre- and post-assessments consisted of self-report measures and a written consumer satisfaction questionnaire. Study III was a randomised controlled study of a brief emotion regulation skills training group. A transdiagnostic sample of adolescents (n = 118) from two child and adolescent psychiatric clinics was randomised to either the intervention group or an active waitlist control. Data were analysed using intent-to-treat principles, with analysis of covariance and exploratory within-group analyses conducted. In Study IV, qualitative interviews were conducted with 10 adolescents and 11 parents regarding their experiences of participating in the emotion regulation skills training. Reflexive thematic analysis was used to analyse the interview data. The findings from Study I demonstrated that the DERS-16 showed acceptable psychometric properties within a child and adolescent psychiatric sample. The measure effectively distinguished between the clinical and community samples in terms of emotion dysregulation. The cluster analysis resulted in a three-cluster solution where the group with highest self-reported emotion dysregulation also were characterised by greater comorbidity, increased risk behaviours, and higher levels of exposure to abuse. Study II indicated that the brief emotion regulation skills training group was feasible, with 87% of participants completing the intervention. Participants generally reported increased knowledge of emotions. Participants showed significant reduction in emotion dysregulation, assessed with Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004) and in alexithymia. The results of Study III showed that participants in the brief emotion regulation skills training group experienced significant reductions in difficulties related to emotional clarity and symptoms of alexithymia compared to those in the control group. There were no significant differences in overall emotion dysregulation, or symptoms of depression and anxiety. In Study IV, qualitative analysis yielded three overarching themes: Parent–Child Processes, Individual Processes, and Group Processes. Participants identified improvements in the parent–child relationship as the primary outcome of the intervention.

The studies in this thesis showed that adolescents in psychiatric care reported greater difficulties with emotion dysregulation than those in the community. The DERS-16 proved suitable for assessing these difficulties, which were linked to higher comorbidity and risk behaviours. A brief emotion regulation skills training, delivered jointly to adolescents and parents, was feasible and associated with reduced difficulties in emotional clarity and symptoms of alexithymia in adolescents. Both adolescents and parents also reported improvements in their relationship following the training.

Tuberculosis (TB) is the leading infectious disease in terms of global mortality and remains a major contributor to poor health worldwide. Infection with Mycobacterium tuberculosis, the causative agent of TB, can result in a wide range of clinical outcomes – from early clearance by the innate immune system to TB infection, subclinical disease and active TB. This variability highlights TB as a disease with a broad clinical spectrum. Although TB is curable with antibiotics, some patients still experience poor outcomes due to relapse or treatment failure. Drug resistance further complicates care, highlighting the need for better tools to monitor treatment and predict outcomes. Epigenetic mechanisms like DNA methylation have shown promise in disease diagnostics and monitoring. Notably, DNA methylation changes can be detected in buccal cells collected via non-invasive mouth swabs, offering a practical approach for clinical sampling. To address the need for improved TB monitoring tools, this thesis investigates whether DNA methylation can serve as a biomarker for TB diagnosis, treatment response and prognosis. The research is driven by the hypothesis that exposure to M. tuberculosis induces epigenetic changes in host immune cells, potentially reflecting disease status and progression. Paper I and II investigate the DNA methylation landscape of host immune cells in the lung compartment and buccal mucosa in healthy controls, individuals exposed to TB and in patients with active TB at the onset of treatment. Notably, TB exposure, TB infection (as indicated by a positive interferongamma release assay) and active TB were associated with distinct DNA methylation signatures in lung and buccal mucosa, reflecting the clinical spectrum of TB. In Paper II, we developed and validated a DNA methylation-based classifier for active TB, comprising seven CpG sites selected for their ability to distinguish TB patients from TB-exposed individuals and healthy controls. Furthermore, the longitudinal study presented in Paper III followed TB patients from the start of treatment through six months of therapy. During treatment, the DNA methylation signature derived from buccal swab samples showed dynamic changes, indicating treatment-associated epigenetic alterations. Using a machine learning approach, we developed a regression model based on nine CpG sites to predict TB symptom severity across multiple populations. Collectively, these findings support the potential of DNA methylation-based biomarkers to aid in TB diagnosis and prognosis. While the hypothesis was partially confirmed (DNA methylation changes were observed in TB exposure and infection) the specificity to M. tuberculosis remains to be fully established. Further longitudinal studies and mechanistic investigations are needed to validate these epigenetic signatures and their clinical utility.

Tree ensembles like random forests and gradient boosting machines are widely used machine learning (ML) models, often outperforming advanced techniques like deep neural networks on structured tabular data tasks. These models also have interpretable (human-understandable) structures that enable stakeholders to trace the decision-making process, making them particularly suitable for use in safety- and security-critical applications where trust in the model’s behaviour is paramount. Despite these advantages, recent work has shown that they are highly vulnerable to adversarial examples: carefully perturbed inputs that elicit misclassifications.

These vulnerabilities are especially concerning as ML continues to permeate domains that are critical to societal functioning. Their seriousness is underscored by legislation such as the recently passed European Union Artificial Intelligence (AI) Act. This act mandates resilience against AI-specific vulnerabilities like evasion attacks caused by adversarial examples targeting ML models at inference time. Measures intended to improve resilience against such evasions, often referred to as hardening, generally involve two strategies: proactive defences, which aim to make models robust (e.g., adversarial re-training), and reactive defences, which focus on detecting and mitigating evasions at inference time. This thesis examines both strategies; it shows that proactive methods like model re-training are ineffective for tree ensembles and consequently advances the state-of-the-art in reactive defences.

In the context of re-training, doubling the training set through targeted data augmentation steps left accuracy largely unchanged. However, robustness, when quantified using formal verification techniques, dropped by 28–82% across two case studies. This indicates that model re-training alone is ineffective for tree ensembles. To address this, we leveraged formal methods to develop Iceman, a prototype system that uses counterexample regions which violate the robustness property to detect evasion attempts. Iceman can detect evasion attacks regardless of the attack generation process without modifying the underlying tree ensemble. It outperforms the current state-of-the-art methods in evasion detection, OC-Score and GROOT. Across four case studies, it improves Matthews Correlation Coefficient scores by 0.20–0.91 and achieves detection speeds 5–115x faster than OC-Score. In addition, it provides alert filtering and prioritisation capabilities with over 98% accuracy to address alert fatigue in intrusion detection systems. However, Iceman’s applicability is limited to scenarios with fixed attacker perturbation budgets, characterised by pre-defined constraints on the input manipulations that an attacker can apply.

To expand this applicability to unconstrained attacker perturbation budgets, we developed an additional system, called Maverick, designed to complement Iceman for a better defensive strategy. Just like Iceman, Maverick does not modify the underlying tree ensemble and can detect evasion attacks regardless of the attack generation process. We prove that Maverick’s core detection mechanism is mathematically equivalent to OC-Score, and present enhancements that achieve 85–563x speedups over OC-Score while maintaining identical detection performance and supporting evasion attack diagnostics with over 93% accuracy.

Rectal cancer is a major global health problem. Advances in surgical technique and preoperative radiotherapy over the past decades have markedly reduced the local recurrence rate, yet overall survival has only been modestly improved. Considerable variation in treatment response persists, reflecting histological and biological heterogeneity. One histological subtype, mucinous adenocarcinoma (MAC), is defined by abundant extracellular mucin and represents roughly 10% of colorectal cancers in Western populations. Compared with non-mucinous adenocarcinoma (NMAC), MAC is more often diagnosed at advanced stage, exhibits greater local invasion, and responds less favorably to radiotherapy and chemotherapy. The mechanisms underlying these differences remain unclear.

The overall aim of this thesis was to determine whether rectal MAC and NMAC should be considered separately in treatment decisions regarding radiotherapy and chemotherapy, to assess potential differences in tumor and stromal protein expression between these subtypes, and to explore whether such protein expression patterns influence prognosis and treatment response.

In paper I, survival outcomes were compared following short-course versus long-course radiotherapy in 289 rectal cancer patients. Patients with NMAC had better survival after short-course than long-course radiotherapy. No difference was seen among patients with MAC. In paper II, the impact of adjuvant chemotherapy on survival was evaluated in 365 patients. Patients with rectal MAC had significant survival benefit from adjuvant chemotherapy, whereas no such benefit was observed in patients with NMAC. In paper III, weighted gene co-expression network analysis (WGCNA) identified mucin-enriched modules associated with the MAC subtype across two independent cohorts, and overlapping genes included MUC1, MUC2, and TCN1. Subsequent Artificial Intelligence (AI)-assisted spatial proteomic analysis of 155 rectal cancers revealed that spatial heterogeneity of these markers, together with tumor nuclear density, was prognostic and predictive of adjuvant chemotherapy benefit, with distinct patterns in MAC and NMAC tumors, suggesting that integrating spatial protein profiling with histological classification may improve prognostic precision and guide individualized treatment strategies in rectal cancer.

Taken together, these studies demonstrate clinically and biologically relevant differences between rectal MAC and NMAC. Integrating histopathology with spatial and molecular biomarker profiling may improve prognostic assessment and enable more personalized treatment strategies.

This dissertation investigates linguistic content in Swedish as a school subject at upper secondary level from a historical perspective. The study focuses on knowledge about language – such as grammar, sociolinguistics and language history – as subject content. Its aim is to contribute to a deeper understanding of the ideas that have shaped the linguistic content of the subject over time.

Drawing on a model that conceptualizes fields within the educational system and using discourse analytical methods, the study examines how linguistic content has been constructed, recontextualized and (de)legitimized by various actors in specific historical contexts.

The dissertation comprises three empirical sub-studies. The first analyzes subject-specific curricula from 1856 to 2025 and the subject’s relationship to the university discipline of Scandinavian languages/Swedish language. It shows a gradual increase in overlap between the school subject and the discipline, resulting from reciprocal influence.

The second sub-study investigates how, and with what arguments, linguistic content has been (de)legitimized by different actors – such as linguists and teachers – in texts published by the Association of Teachers of Swedish. The findings show that arguments for and against specific content are shaped by the historical context, reflecting continuity and change in society and the educational system as well as linguistic and subject-specific educational research.

The third sub-study combines analyses of legitimation and dialogicity to examine which arguments and voices from the subject-related discussions are recontextualized in subject-specific curricula. The results reveal a genre shift from reflective texts to more prescriptive ones, suggesting a declining trust in the teacher’s pedagogic and subject-specific expertise.

Together, the findings contribute to a deeper understanding of how ideas about linguistic knowledge content in the subject have been shaped by institutional, ideological and disciplinary forces over time.

This study explores how student-generated comic books can provide insight into students? meaning making about evolution. Despite the complexity of evolution, previous research indicates that students can grasp concepts like natural selection through age-appropriate activities. In this study, we followed a group of Grade 4 students over seven weeks as they engaged in meaning making about evolution. As part of the lessons, the students read comics about evolution, learnt about comic conventions, and at the end of the unit, they created their own comics to represent their understanding of evolution. Building on multimodal theory rooted in social semiotics, we have analysed 14 student comics, focusing on what evolutionary principles are shown in the texts and the semiotic resources used to express these. Findings reveal that all targeted evolutionary principles (variation, heredity, and selection) were represented in the comics, though not all comics included every concept. Evolutionary ideas were expressed multimodally, with common patterns showing evolution at a population level. The sequentiality of the comic format facilitated the depiction of adaptation through natural selection. This study provides insights into the potential of using comics as a tool for students to make meaning and communicate their knowledge about evolutionary concepts.

Many of the most widespread neurodegenerative diseases, such as Alzheimer’s disease and Parkinson’s disease, stem from protein misfolding and aggregation, in part as a consequence of an imbalance in the protein homeostasis (proteostasis). These diseases are devastating for the affected patients, their relatives, and our society. Today, these diseases are notoriously hard to cure or even treat.

Molecular chaperones make up the cellular machinery that handle unfolded, misfolded, and aggregated proteins. Throughout evolution chaperones have been a prerequisite for the existence of complex proteins. This has made chaperones essential for biological life as we know it.

Utilizing chaperones as a pharmaceutical drug is a novel treatment approach for neurodegenerative disease. By adding molecular chaperones to the system, one could potentially improve the cells’ ability to handle the cellular stress that comes because of the imbalance of the proteostasis. Today there are several chaperone candidates that are being explored for this purpose.

HSP10 is an essential co-chaperone to HSP60. However, HSP10 has also been found to act alone. HSP10 has been shown to inhibit autoimmune activity. The co-chaperone is systemically overexpressed in the early stages of pregnancy. The overabundance of HSP10 in the brain compared to HSP60 and association with neurodegenerative disease suggests that the co-chaperone could have more functions than previously documented.

This thesis investigates how HSP10 interacts with several different amyloid proteins during amyloid fibrillation both in an in vitro and in an in vivo setting. For in vitro studies, interactions between recombinant proteins were scrutinized using a range of biophysical measurement techniques, e.g. aggregation kinetics, transmission electron microscopy, small angle X-ray scattering among others. The in vivo model Drosophila melanogaster was utilized to monitor how the combination of HSP10 and aggregation prone protein expression affects a complex organism.

Aggregation kinetics of the neurodegenerative peptide Aβ1-42, associated with Alzheimer´s disease, was performed in the presence of different concentrations of HSP10. Aggregation inhibition of Aβ1-42 was evident when HSP10 was present at high concentrations. However, low concentration of HSP10 in contrast accelerated the aggregation of Aβ1-42. This was followed by aggregation kinetics with α-synuclein, a protein associated with Parkinson´s disease. The presence of HSP10 at high concentration yielded a complete inhibition of α- synuclein aggregation. However, no acceleration of α-synuclein fibrillation was seen when the concentration was lowered. Further investigation of aggregation kinetics with the familial disease associated α-synuclein mutant A30P revealed the dual property of inhibition at high concentration of HSP10 and acceleration at lower concentration of HSP10.

To investigate how HSP10 would act in vivo, we used Drosophila melanogaster (Drosophila) as a model system. HSP10 was overexpressed in the neurons of Drosophila with and without the presence Aβ1-40 and Aβ1-42 and the lifespan and activity of the flies were monitored and were compared with controls. Overexpression of HSP10 in wild type flies without Aβ1-42 significantly prolonged the lifespan of Drosophila compared to control flies demonstrating that HSP10 rendered a protective effect against aging. Overexpression of Aβ1-42 decreased the lifespan of Drosophila. Surprisingly, the overexpression of both HSP10 and Aβ1-42 lead to an even shorter lifespan than expression of Aβ1-42 alone, suggesting an accelerated fibril formation process with enhanced neurotoxicity.

To investigate if the dual property of concentration dependent inhibition and acceleration of amyloid formation was unique to HSP10, fibril formation studies of Aβ1-40 and Aβ1-42 was conducted in the presence of the molecular chaperones αB-crystallin (from the small HSP chaperone family) and DNAJB1 (from the HSP40 chaperone family). DNAJB1 and αB-crystallin are classified as canonical ATP-independent holdases. Both DNAJB1 and αB-crystallin chaperones showed a clear inhibition of amyloid fibril formation when they were present at high concentration while they accelerated the fibril formation when the chaperone concentrations were lowered. Hence, we concluded that chaperone acceleration of amyloid formation appeared to be a general chaperone mechanism of holdases and not a unique feature for HSP10.

Herein, we have investigated fundamental mechanisms of molecular chaperones in the context of modulation of amyloid fibril formation. Our results uncovered a new chaperone property: acceleration of amyloid fibril formation and provided valuable insights into the molecular mechanisms of chaperones. This thesis contributed with crucial knowledge needed for development of molecular chaperones as pharmaceutical agents targeting neurodegenerative disease.

Modeling and simulation are used to solve real-world problems safely and efficiently. For example, we can conduct simulations and analyze the results by constructing digital models of cyber-physical systems to make informed decisions. The field of Modeling and Simulation has recently grown and is tackling systems with increased complexity and size. Hence, modeling and simulating cyber-physical systems are becoming increasingly complex, requiring traditional modeling languages and tools to adapt. Modelica is an open-standard declarative equation-based object-oriented language used to model various systems. Existing tools allow modelers to model and simulate large, complex systems. However, the language and existing tooling cannot express some concepts, such as radical structural changes to the components or the behavior of systems during simulation. We propose several language extensions to support modeling variable-structure systems (VSS), that is, systems in which the system’s structure may radically change during simulation. To achieve these goals, we have developed a new modeling framework that supports the language itself alongside several extensions. The proposed extensions can handle explicit and implicit modeling of VSS by new operators and, consequently, new semantics for the language. Furthermore, we introduce and investigate additional features in terms of operators and semantics to aid VSS modeling and simulation.

One such extension is dynamic-overconstrained connectors (DOCC), which are helpful, for instance, when modeling AC transmission systems, particularly large ones. In the thesis we provide a technique to better handle such models by relaxing existing modeling constraints. Another extension is the inclusion of a experimental operator for inline integration and investigating the practical benefits of its introduction. The environment has been used to develop a new language that merges aspects of Context-Oriented Programming and declarative equation-based modeling, utilizing the introduced primitives. Furthermore, we have also validated the framework by using it to simulate relvant and recent climate models.

Explicit VSS modeling is based on extensions, which provide the possibility of switching configuration at runtime between model definitions resolved at compile time. The implicit modeling supports reconfiguration during runtime via recompilation. A just-in-time compiler was implemented to handle the new semantics using the symbolic-numeric programming language Julia. We investigate the performance of our new framework and compare it to existing state-of-the-art tooling on models with over 100,000 equations.

The results show that the extensions, framework and methods are viable for simulating both regular models and models with structural variability. We demonstrate that the Modelica language can be extended to support systems with variable structures by providing additional operators and enhanced runtime system support.

The European Green Deal (EGD) was introduced as a transformative policy agenda for sustainability aiming to make Europe the first climate-neutral continent in the world. While previous research has studied its transformative potential, there is no comprehensive review of that research. We provide an integrative review of discursive policy analyses of the EGD, arguing that it is pivotal that policymakers understand what transformational potentials the agenda has—along with the obstacles to realize it in transformative ways that are inherent in its formulations. We discuss the transformative potential of the EGD, as presented in the 26 articles included in our review, using a theoretical framework of “empty signifiers” and “ivy discourses”. We conclude that the agenda is characterized by a certain amount of transformational discourse. However, its transformative impact risks being limited as it is enmeshed in assumptions of de-coupling and international relations remain tainted by colonial power dynamics.

Syftet med föreliggande studie ar att undersöka relationen mellan tendens till moraliskt disengagemang i mobbningssituationer, kon och mobbning. Moraliskt disengagemang refererar till sociokognitiva processer som rättfärdigar, bortrationaliserar eller bagatelliserar inhumana beteenden och dess negativa konsekvenser och som leder till att människor kan begå inhumana handlingar utan att kanna skuld eller anger. I denna studie undersöktes moraliskt disengagemang som en generell variabel men aven uppdelat i dess fyra grundkategorier (kognitiv omkonstruering, minimera det egna ansvaret, bortse från eller förvränga konsekvenserna av skadligt beteende och offerattribuering). Trehundrasjuttio mellanstadieelever besvarade en enkät i deras ordinarie klassrum.

Resultatet visade mobbning korrelerade signifikant och positivt med moraliskt disengagemang, kognitiv omstrukturering, minimering av ansvar, minimering av konsekvenser. Moraliskt disengagemang som generell variabel men aven samtliga fyra grundkategorier av moraliskt disengagemang var dessutom högre hos pojkar än hos flickor. Vidare visade resultatet att moraliskt disengagemang, kognitiv omkonstruering, minimerande av konsekvenser och offerattribuering var signifikant högre hos mobbare jämfört med offer och oinvolverade. Mobbare uppvisade aven högre grad av att minimera ansvar jämfört med oinvolverade. Slutligen visade resultatet att offer uttryckte en högre grad av kognitiv omkonstruering an oinvolverade. Sammantaget föreslår resultatet vikten av att göra elever uppmärksamma på moraliskt disengagemang och dess olika grundkategorier och att försöka minska förekomsten av dessa processer som en del av skolans antimobbningsarbete.

Machine learning is a promising technique for angle-of-arrival (AOA) estimation of waves impinging a sensor array. However, the majority of the methods proposed so far only consider a known, fixed number of impinging waves, i.e., a fixed number of sources (NOS). This paper proposes a machine-learning-based estimator designed for the case when the NOS is variable and hence unknown a priori. The proposed estimator comprises a framework of single-label classifiers. Each classifier predicts if waves are present within certain randomly selected segments of the array’s field of view (FOV), resulting from discretising the FOV with a certain (FOV) resolution. The classifiers’ predictions are combined into a probabilistic angle spectrum, whereupon the NOS and the AOAs are estimated jointly by applying a probability threshold whose optimal level is learned from data. The estimator’s performance is assessed using a new performance metric: the joint AOA estimation success rate. Numerical simulations show that for low SNR (−10 dB), a low FOV resolution (2°) yields a higher success rate than a high resolution (1°), whereas the opposite applies for mid (0 dB) and high (10 dB) SNRs. In nearly all simulations, except one at low SNR and a high FOV resolution, the proposed estimator outperforms the MUSIC algorithm if the maximum allowed AOA estimation error is approximately equal to (or larger than) the FOV resolution.

This dissertation explores how very young children (1–3 years old) in Swedish preschools encounter and engage with science through everyday play and interactions. The study aims to generate knowledge about how emergent science learning takes shape in early childhood education and how children’s actions, individually and together with peers and preschool teachers, can be understood as scientific meaning-making.

The theoretical framework is grounded in symbolic interactionism, highlighting how meaning emerges through social interaction and the use of social objects. Methodologically, the study draws on a constructivist grounded theory approach, based on ethnographic fieldwork including video observations, field notes, and informal conversations in several preschool settings.

The analysis identifies recurring patterns in children’s meaning-making: exploring independently (en-skapar), co-creating with peers (kamrat-skapar), and jointly creating with teachers (barn-pedagog-skapar). Central strategies in these processes are repetition, variation, and synthesis, which support children’s bodily, cognitive, and conceptual engagement with scientific phenomena such as gravity, friction, and magnetism. The study also shows that teachers play a crucial role: when they respond to and extend children’s initiatives, spontaneous and meaningful science learning opportunities arise, whereas children’s explorations risk being overlooked when teachers are absent or inattentive.

The findings contribute to the growing field of research on emergent science by providing insights into how very young children’s everyday actions can be seen as foundational scientific practices. Pedagogical implications include the need for attentive and responsive teaching, access to rich environments and materials, and the use of scientific language to gradually deepen children’s conceptual understanding. The study underscores preschool as a key arena for fostering curiosity, wonder, and long-term interest in science, not by transmitting facts but by creating conditions for children’s own explorations in dialogue with peers, materials, and educators.

Objectives: Research on complex posttraumatic stress disorder (CPTSD) among individuals with refugee backgrounds is limited, particularly in adolescents, and its validity in this group remains underexplored. The aim of this study was to assess the prevalence and discriminant validity of posttraumatic stress disorder (PTSD) and CPTSD, as well as the risk factors for CPTSD, in a community sample of adolescents with refugee backgrounds residing in Sweden.

Methods: The study included 296 adolescents with refugee backgrounds recruited nationwide in Sweden. Participants were aged 12-25 years, 45.3% were female, and 23.7% had unaccompanied status. Probable diagnoses were evaluated according to DSM-5 and ICD-11 criteria using questionnaires. Latent class analysis was employed to examine the discriminant validity of PTSD and CPTSD, while logistic regression analysis was used to explore risk factors for CPTSD.

Results: The findings indicated that 24.1% had a probable diagnosis of PTSD according to the DSM-5. For ICD-11, the equivalent proportions were 7.1% for PTSD and 10.8% for CPTSD. The probable diagnostic rates for DSM-5 PTSD were significantly higher than ICD-11 PTSD and CPTSD. Latent class analysis identified three distinct classes: Low symptoms (46.9%), PTSD (29.6%), and CPTSD (23.6%). Compared to the PTSD class, membership in the CPTSD class was predicted by exposure to more types of violence and child maltreatment. It was also associated with higher posttraumatic stress symptoms, worse general functioning, poorer mental well-being, increased suicidal ideation, more treatment-seeking behavior and greater comorbidity.

Conclusions: This study found a high prevalence of PTSD and CPTSD among adolescents with refugee backgrounds living in Sweden. Distinct classes aligned with the ICD-11 formulation of PTSD and CPTSD were identified, with exposure to more types of violence and child maltreatment emerging as key risk factors for CPTSD. Key messages • Adolescents with refugee backgrounds living in Sweden may be highly affected by PTSD and CPTSD. • The ICD-11 distinction between PTSD and CPTSD is likely valid for adolescents with refugee backgrounds. Exposure to more types of violence and child maltreatment may be risk factors for CPTSD.

Progressing with climate adaptation in dense urban built environments has proved challenging. Making adaptation multifunctional has been proposed as a promising approach, allowing effective use of limited urban space by simultaneously meeting important societal needs. However, there is still uncertainty about the extent to which multifunctionality is actively planned, designed, and implemented in urban areas and what factors influence implementation uptake. This paper analytically unpacks the practice and prospects of multifunctional adaptation in urban built environments, illustrating empirical lessons from ten Swedish cities. We conclude that multifunctional adaptation in practical examples tends to focus more on manifesting and effectively using existing green space for multiple functions instead of undoing and transforming traditional grey urban environments to enhance greening and adding new green qualities in the urban landscape. Moreover, city size proves important when considering both motivations and implementation uptake. Larger cities motivate multifunctional adaptation by political expectations and lack of available space, whereas in smaller cities motivations include their contribution to urban attractiveness and public acceptance of adaptation measures. Many factors identified in previous studies as influencing implementation uptake also prove valid for multifunctional urban built environment adaptation. However, there is evidence of substantial size-related differences in the analysis, with some factors being more manageable in smaller and others in larger cities. This suggests the importance of analytically distinguishing city size for the sake of research accuracy, policy evaluation and design, to be able to make the most of ongoing multifunctional adaptation in dense and complex urban built environments.

Background: Cognitive problems are common among cardiac arrest survivors, but current knowledge is largely based on performance-based tests, which do not capture survivors perceived cognitive function. To date, there is no specific self-report instrument designed to assess cognition in this population.

Aim: The overall aim of this thesis was to expand knowledge about cardiac arrest survivors’ cognitive function to enhance the care and assessment.

Methods: This thesis comprises four studies employing different research designs: a qualitative interview study (I), two quantitative retrospective registry studies (II, III), and an instrument development study (IV). In study I, semi-structured interviews were conducted with registered nurses (n = 19) included by purposeful sampling. The nurses were experienced in cardiovascular care, including the management of cardiac arrest and participation in post-cardiac arrest follow-up conversations. Data was analysed with a phenomenological approach. Studies II and III utilised data from the Swedish Register of Cardiopulmonary Resuscitation. In study II, adult cardiac arrest survivors with clinician-reported Cerebral Performance Category (CPC) scores of 1 or 2 and self-reported health status and psychological distress at follow-up 3–6 months post-arrest were included (n = 2058). Self-reported health was compared between the groups of survivors with CPC 1 and CPC 2, and differences between the groups were analysed with regression models. Study III included adult cardiac arrest survivors with self-reported cognitive function, health status, psychological distress, and overall life satisfaction at follow-up, 3–6 months post-arrest (n = 1254). In this study, the association between self-reported cognitive function and the outcomes were explored using binary logistic regression models. Study IV involved five stages: 1) item generation, 2) Delphi study, 3) Content Validity Index, 4) cognitive interviewing, and 5) final revision and scoring. Stage 1 included a conceptual framework of cognitive function, literature search for instruments with self-reports of cognition, mapping items to the cognitive domains and formulating the items. Stage 2 included two Delphi rounds, were three expert groups (adult cardiac arrest survivors, adult family members and healthcare professionals), evaluated the relevance of the items. In stage 3, content validity was evaluated from an expert perspective by healthcare professionals with experience in post-cardiac arrest care and/or neurorehabilitation. For the cognitive interviews in stage 4, adult cardiac arrest survivors were interviewed regarding the response processes. Final revisions of the instrument were done in stage 5. 

Findings: The findings showed that registered nurses perceived the assessment of cognitive impairment after cardiac arrest challenges, especially regarding milder deficits. They had various strategies to identify cognitive impairments, but these did not routinely include standardised instruments. Additionally, nurses perceived their assessments of survivors functioning were sometimes misinterpreted. Furthermore, cardiac arrest survivors with a moderate cerebral disability and those with self-reported cognitive decline showed a significantly worse self-reported health, higher psychological distress and low life satisfaction compared to those with good cerebral performance and no self-reported cognitive decline. The development of the Self-Reported Cognition – Cardiac Arrest (SeReCo-CA) resulted in a 29-item instrument, encompassing the cognitive domains of learning and memory, language, complex attention, executive function and perceptual-motor function.

Conclusion: This thesis underscores the importance of assessing cognitive function in cardiac arrest survivors, showing that perceived cognitive decline, even when mild, is linked to poorer health, psychological distress, and reduced life satisfaction. Although screening is recommended in post-resuscitation care guidelines, it remains underutilized, partly due to challenges in detecting subtle impairments. The newly developed SeReCo-CA instrument offers a comprehensive self-report tool that may help bridge this gap, though further psychometric validation is needed.

Aim Self-reported cognitive function has been described as an important complement to performance-based measurements but has seldom been investigated in cardiac arrest (CA) survivors. Therefore, the aim was to describe self-reported cognitive function and its association with health status, psychological distress, and life satisfaction. Methods This study utilised data from the Swedish Register of Cardiopulmonary Resuscitation (2018–2021), registered 3–6 months post-CA. Cognitive function was assessed by a single question: “How do you experience your memory, concentration, and/or planning abilities today compared to before the cardiac arrest?”. Health status was measured using the EQ VAS, psychological distress with the Hospital Anxiety and Depression Scale, and overall life satisfaction with the Life Satisfaction checklist. Data were analysed using binary logistic regression. Results Among 4026 identified survivors, 1254 fulfilled the inclusion criteria. The mean age was 65.9 years (SD = 13.4) and 31.7% were female. Self-reported cognitive function among survivors was reported as: ‘Much worse’ by 3.1%, ‘Worse’ by 23.8%, ‘Unchanged’ by 68.3%, ‘Better’ by 3.3%, and ‘Much better’ by 1.5%. Declined cognitive function was associated with lower health status (OR = 2.76, 95% CI = 2.09–3.64), symptoms of anxiety (OR = 3.84, 95% CI = 2.80–5.24) and depression (OR = 4.52, 95% CI = 3.22–6.32), and being dissatisfied with overall life (OR = 2.74, 95% CI = 2.11–3.54). These associations remained significant after age, sex, place of CA, aetiology, initial rhythm, initial witnessed status, and cerebral performance were controlled. Conclusions Survivors experiencing declined cognitive function post-CA are at a higher risk of poorer health status, increased psychological distress, and reduced life satisfaction, and these risks should be acknowledged by healthcare professionals.

Children with Speech, Language and Communication Needs (SLCN) constitute a broad and diverse group. In this thesis, the term is used to refer to children with Developmental Language Disorder (DLD) or phonologically based Speech Sound Disorder (SSD), as these conditions fall within the scope of the included studies.

Children with SLCN often face challenges in everyday life due to their difficulties, for example in terms of a negative impact on their interaction with peers. Accordingly, improved everyday communication and social inclusion are commonly reported by caregivers as primary goals for treatment outcomes. Speech-Language Pathologists (SLPs) have also emphasised that it is important that their services target abilities relevant to the children’s daily communication. However, SLPs in Sweden traditionally meet children with SLCN in clinical settings within the healthcare system, rather than in the children’s everyday environments. Thus, SLPs often have limited direct access to children’s everyday abilities and challenges, and must therefore rely on other sources to obtain this information. Video recordings from home environments may offer an efficient means of accessing such information. While video recordings are currently used within certain parental programs in Swedish SLP services, their use outside of these programs remains largely unexplored.

The overarching aim of this thesis is to advance knowledge about SLP services provided to preschool children with SLCN in Sweden, with a focus on how these services align with the children’s everyday lives. Furthermore, the thesis seeks to contribute insights into the use of video recordings as part of clinical service delivery. The thesis comprises four studies, each addressing this aim from different perspectives.

Study I was an investigation of how SLPs in Sweden align their clinical practices to the everyday language and communication abilities of children with (D)LD, and their use of video recordings as part of clinical practices. A web-based questionnaire was answered by 163 SLPs. The findings suggested that information about children's everyday abilities was primarily collected through caregiver reports and observations of the children at the clinic. The use of video recordings for clinical purposes was very limited.

In Study II, the aim was to extend current knowledge about possibilities and challenges encountered by Swedish SLPs in targeting aspects of everyday language and communication in children with DLD. Unstructured focus-group discussions were conducted with a total of 15 SLPs participating. The results demonstrated that the SLPs viewed themselves as detached from the children’s everyday context. Collaboration with caregivers and (pre)school staff was emphasised for aligning services with children’s everyday life. However, the SLPs often experienced a gap between different stakeholders, and this was a challenge for providing the most appropriate care. The services that the SLPs could offer were largely regulated by local routines, and children requiring more extensive care were sometimes disfavoured by these routines.

Study III was an exploration of how a treatment approach using video recordings from children’s home environment might supplement regular SLP treatment for preschool children with SLCN. Four children and their caregivers first received regular treatment at their SLP clinic, followed by a new video-based treatment approach. The video-based treatment invited the caregivers to reflect on their child’s language use and positioned them as experts on their child. The video recordings provided the SLP with insights into the child’s everyday abilities and served as a foundation for caregiver coaching and collaborative discussions based on real-life situations.

In Study IV the aim was to investigate how the video-based treatment, explored in Study III, was perceived by caregivers to children with SLCN, in relation to regular SLP treatment. The caregivers were interviewed at three occasions during the treatment process. The videobased approach was described to enhance caregivers’ sense of involvement, although the role of the SLP as an expert remained central. Caregivers appreciated the concrete methods used in regular treatment but found the video recordings helpful in aligning support with the children’s everyday abilities. Video-based coaching was perceived as more tangible than regular treatment, due to its grounding in real-life situations, but newly introduced practices were difficult to sustain.

The main findings of this thesis suggest that the alignment of SLP services in Sweden with the everyday lives of children with SLCN largely depends on the abilities and circumstances of the children’s caregivers. Although video recordings are rarely used routinely in clinical practice, they hold significant potential to provide SLPs with important insights into children’s everyday abilities and to serve as a foundation for caregiver coaching based on real-life situations. Integrating video from children’s home environments into existing SLP routines could thus be a simple way to enhance the alignment of the services with the daily lives of children with SLCN and their families.

The Confederation of Swedish Enterprise’s annual ranking of Swedish municipalities’ business climates has become an influential tool in local strategic planning, despite recurring criticism. One key driver of the ranking’s impact is the extensive coverage it receives in the local press. This study examines how local newspapers report on the ranking and whether their coverage fulfils journalism’s democratic functions: informing citizens, scrutinising power, and facilitating public debate. The investigation draws on a content analysis of 570 articles from 2008, 2014, and 2024, complemented by theories of political communication and democratic ideals. The findings show that local media often relay the ranking with limited critical distance, meaning journalism acts more as an amplifier than as democracy’s watchdog. Concepts such as agenda-setting, priming, and hegemony are employed to demonstrate how the media help legitimise a narrow definition of what constitutes a “good” business climate. A normative assessment based on three models of democracy—elite, institutional, and deliberative—suggests that the media’s handling of the ranking falls short of realising its full democratic potential.

Denna artikel följer transformationen av svensk asyl-, migrations- och integrationspolitik under de senaste tre decennierna och identifierar tre centrala skiften: från permanenta till tillfälliga uppehållstillstånd, från integration till återvändandefokuserad politik, samt från Redistributive Discourse (RED) till moraliserande diskurser.

Analysen är indelad i tre tidsperioder: 1990-talet, åren 2000–2022 (fram till Tidöavtalet), samt perioden 2022 och framåt (efter Tidöavtalet). Den belyser den gradvisa nedmonteringen av rättighetsbaserade ansatser samt framväxten av villkorlighet, övervakning och exkludering.

Tidöavtalet markerar ett paradigmskifte genom att institutionalisera en Moral Underclass Discourse (MUD), där migranter inte bara problematiseras ekonomiskt utan även moraliskt, genom mekanismer som deportabilitet och permanent tillstånd av villkorad tillhörighet.

Artikeln tar sin utgångspunkt i det engelska begreppet crimmigration och visar hur samman­smältningen av kriminal- och migrationslagstiftning intensifierar den rättsliga, ekonomiska och emotionella utsattheten för asylsökande.

Avslutningsvis ifrågasätts effektiviteten i avskräck­ningsbaserad politik och den dominerande diskursen om grundorsaker (eng. root causes) utmanas. I stället hävdas att strukturella ojämlikheter, förankrade i kolonialt arv och samtida nyliberal kapitalism, tillsammans med en bestående efterfrågan på arbetskraft i det globala nord, är de främsta drivkrafterna bakom migration.

Artikeln uppmanar till en omprövning av både utgångspunkterna och målen för migrationspolitiken, grundad i en kritisk förståelse av den globala politiska ekonomin.