Time-series forecasting plays a crucial role in supporting data-drivendecisions across domains like energy systems, finance, and healthcare.Despite their strong predictive power, many deep learning models arestill viewed as "black-boxes", making it difficult to understand howthey arrive at their predictions. This lack of transparency often limits their broader adoption in real-world applications. To address thischallenge, this thesis proposes an interpretable, prototype-based deeplearning framework for multivariate time-series forecasting that integrates both numerical and textual modalities to enhance predictiveaccuracy and transparency.

At the core of the proposed framework lies the Prototype Similarity Network (ProSeNet), which learns a set of representative temporal patterns called prototypes that serve as interpretable anchors formodel reasoning. Each forecast is explained in terms of similarity tothese learned prototypes, providing clear insights into the model’sdecision process. To further enhance contextual understanding, theframework is extended into a multimodal architecture where textualembeddings derived from pretrained language models are fused withtemporal features, enabling the model to reason jointly over structured and unstructured data.

Comprehensive experiments demonstrate that the proposed approachachieves strong predictive performance while maintaining interpretability. The prototype-based design allows users to trace predictions backto meaningful temporal patterns, and the integration of textual embeddings significantly improves contextual reasoning and accuracy.This work contributes a transparent, context-aware forecasting framework that bridges interpretable representation learning and multimodal reasoning advancing the development of trustworthy and explainable AI systems for time-dependent applications.

Background: Substance use disorder is a complex condition that affects an

individual's physical, mental and social well-being. The diagnosis presents

significant nursing challenges and places high demands on health care

services. Aim: To explore patients with substance use disorder experiences

of interactions with healthcare professionals within the healthcare system.

Method: A general literature review with an inductive approach, based on

12 scientific articles that were critically appraised and analyzed. Results:

Findings, presented in two themes, describe both negative and positive

healthcare experiences. Patients reported stigmatizing attitudes,

discrimination, and oppression, leading to feelings of powerlessness and

dehumanization. Positive experiences were reported where interactions were

described as open-minded and respectful, which increased the patients' trust

in healthcare staff and motivation for engagement and participation.

Conclusion: Healthcare encounters significantly impact patients' health and

well-being. It's essential for nurses to adopt a person-centered approach and

to treat each patient with respect and goodwill while adhering to the ethical

code. From an educational perspective, it is important to develop and

deepen knowledge about substance use disorder and how to appropriately

engage with this patient group. Finally, further research is needed to

enhance understanding of substance use disorder and to highlight the

importance of healthcare encounters.

Artificial intelligence (AI) has become a pervasive technological enabler reshaping business practices and entrepreneurial activity. While its potential to optimize decision-making and foster innovation is widely acknowledged, its implications for entrepreneurs remain contested. The aim of this paper is to explore how entrepreneurs can use AI effectively and responsibly in business decision-making by addressing three research questions: (1) What level of AI knowledge do entrepreneurs need to make responsible and informed decisions? (2) How do different types of AI tools support or hinder entrepreneurial decision-making? (3) What ethical challenges do entrepreneurs face when using AI, and how can they ensure transparency and accountability? Drawing on semi-structured interviews with ten entrepreneurs and complementary case materials from three research projects conducted the study applies thematic analysis to examine entrepreneurs’ perceptions, practices, and dilemmas. The findings show that entrepreneurs require at least basic literacy in AI concepts to avoid misuse and over-reliance. Generative AI was valued for creativity and efficiency but distrusted for decision-making, while supervised machine learning was perceived as a more reliable tool for prediction and analysis. Ethical concerns were prominent, including transparency gaps, accountability, data bias, and regulatory uncertainty. The study concludes that entrepreneurs should adopt hybrid models where human judgment and AI complement one another, with responsibility and critical reflection resting firmly on the entrepreneur. Implications are outlined for entrepreneurial education, managerial practice, and policy regulation.

Sex education has been mandatory in Swedish schools since the 1950s. For several decades, it has been characterized by a narrow perspective on sexuality related issues and has been limited to a more biomedical approach. In 2022, the governing documents of the Swedish school system were revised, and the curriculum clarified a healthpromoting approach to sex education, which was also renamed to "Sexual consent, and Relationships."The aim of this study was to describe educators’ perceptions of the subject of Sexual consent, and Relationships following the changes to the curriculum. The theoretical framework of the thesis is based on a sociocultural perspective on learning and the concept of Sense of Coherence (SOC). The study has a hermeneutic approach, and the methodological starting point has been abductive.Ten semi-structured interviews were conducted and analyzed using qualitative content analysis. The results of the study showed an increased awareness of the subject area. At the same time, educators reported a knowledge gap regarding sexual consent, and relationships, leading to difficulties in approaching sexuality related topics with students. The results also showed that educators view communication and collaboration with other internal professionals such as student health teams and school nurses as wellas with external actors, as important factors for learning in relation to sexual consent, and relationships.

Background: Existential suffering is a condition that can often be associated with palliative care. It is a subjective form of suffering that is experienced in different ways depending on the individual's past experiences. It is part of the nurse's professional role to identify and support the patient and alleviate suffering caused by existential issues. Aim: The aim was to identify nurses’ experiences of supporting patients with existential suffering within palliative inpatient care. Method: A general literature study based on eight scientific articles, four were quantitative studies and four were qualitative studies. Results: The results identified four categories based on the nurse's experiences of meeting patients with palliative care and existential suffering: To be emotionally affected in the meeting with the patient, That experience and education on existential suffering is significant in the meeting with the patient, To adjust the interaction based on the patient's unique needs, That existential suffering is not prioritised in the meeting with the patient. Conclusion: More research is required in how nurses should support and care for patients in palliative care with existential suffering to promote good care and improve the quality of life of patients at the end of their life. 

This thesis investigates the challenges associated with task allocation and motion planning in dynamic and complex environments involving fleets of mobile robots. The primary objective is to coordinate task allocation and trajectory planning in a manner that promotes balanced workload distribution, formulated as the minimization of the maximum operational cost incurred by any individual robot. A hybrid planning framework is proposed in which centralized task allocation and global path planning are performed under static assumptions, while execution-level feasibility is maintained through local trajectory refinement. Task allocation is formulated using a permutation-matrix representation and explored using multiple solution strategies, including deterministic annealing with Potts neurons, stochastic simulated annealing, and heuristic graph-based methods. The cost matrix, derived from global path planners and representing path length and/or traversal time, enables a systematic comparison of these approaches in terms of solution quality, computational complexity, and scalability. The results indicate that deterministic annealing can produce high-quality, balanced task allocations for small- to mediumscale problem instances, while heuristic methods offer improved robustness and computational efficiency in larger or time-critical scenarios. At the motion-planning level, the framework adapts the CHOMP (Covariant Hamiltonian Optimization for Motion Planning) algorithm to account for the kinematic constraints of non-holonomic wheeled mobile robots. Rather than serving as a standalone global planner, the modified CHOMP formulation is used as a local trajectory refinement mechanism, enabling collision avoidance and feasibility preservation during execution. Experimental and simulation results demonstrate that this approach improves trajectory smoothness and feasibility in moderately dynamic environments, while also highlighting limitations related to scalability and sensitivity to initialization. Overall, this thesis presents an integrated task and motion planning framework that emphasizes structured problem formulation and systematic trade-off analysis rather than universal optimality. By explicitly examining the conditions under which different allocation and motion-planning strategies are effective, the work contributes practical insights into multi-robot coordination and supports the informed deployment of autonomous robotic systems in industrial and logistics automation.

Many public healthcare datasets suffer from severe missing data, resulting in high sparsity. Reports indicate that in some public sources, over 80% of the data contain missing values. Ignoring these missing entries can significantly impact downstream clinical predictions, particularly in data-driven models. Consequently, imputing missing values is essential for effective healthcare data analysis and mining. Meanwhile, healthcare datasets inherently possess a graph structure that captures the complex relationships among patients, clinical measurements, and diseases. Existing imputation methods exhibit limitations. Most graph-based models, such as the discriminative imputation model GRIN, are restricted to homogeneous graphs with a single node type, limiting their ability to capture cross-type interactions, and do not provide any uncertainty estimate of the imputed value. Generative diffusion-based models, on the other hand, like CSDI, achieve strong performance, but they do not incorporate explicit graph structure to leverage relational dependencies in the data. To address this problem, we propose a novel Variational Graph Autoencoder (VGAE)-based model that integrates Chebyshev graph convolution, spatial attention, and cross-attention fusion to combine patient- and measurement-based graph views, thereby capturing richer relational information. Experiments conducted on the MIMIC-IV dataset, which comprises over 36,000 patients, eight clinical measurements, and 48 time steps, demonstrate that our model outperforms several baseline methods, such as  Mean, KNN, MICE, MF and Brits, and achieves very close imputation error to CSDI and GRIN. The fused cross-attention approach achieves lower imputation errors than using either graph view individually. Furthermore, the VGAE framework enables uncertainty estimation for each imputed value, which is particularly beneficial in clinical decision-making. These results highlight that leveraging multiple node perspectives through cross-attention offers a promising direction for improving healthcare data imputation. 

Background: Multiple Sclerosis (MS) is a neurological disease in which the nerve pathway is damaged and the myelin breaks down. The breakdown of myelin affects the nerve signals from the brain to the spinal cord, and the immune system becomes disrupted. Suffering is an individual feeling with only the patient able to express their experience. Through inner strength, meaning in life and supportive functions, suffering can be alleviated. Purpose: To describe patients’ experience of alleviating suffering in Multiple Sclerosis. Method: A general study with a qualitative method. A structured search in CINAHL, PubMed and PsycINFO resulted in eight current articles. Result: Five categories were identified: Suffering and acceptance, Suffering and meaningful activities. Suffering and family, Suffering and information, Suffering and the society. Conclusion: By being able to handle suffering instills a sense of hope and strengthens the patient’s self-esteem. Accepting illness and adopting a positive attitude provides inspiration to find new meaning in life. The study has shown that distinct information and communication are of significant importance to be able to handle suffering for the patient themself and contributing to better family support. Patients experience that distinct information and communication are important for society to be able to contribute to alleviating suffering. 

Background: Myocardial infarction represents a turning point for patients in considering changing their lifestyle habits. Physical activity is recommended as a secondary prevention for all patients who have had a myocardial infarction. The nurse’s responsibility is to identify each patient’s individual needs and conditions for physical activity. Aim: The literature review aimed to illustrate patients’ experiences of physical activity after a myocardial infarction. Method: A general literature review was conducted, including ten qualitative articles. The articles were processed according to Popenoe’s analytical method. Result: Three categories were identified: a new existence, the balance between uncertainty and agency, and the safety of the community. Patients experienced a changed existence, which affected their conditions for physical activity after a myocardial infarction, as both body and heart were perceived as changed. Patients described that physical activity after a myocardial infarction could cause concern about overstraining the heart muscle. A sense of security derived from close relatives, other patients, and healthcare professionals fostered increased commitment to physical activity. Conclusion: Patients express a need for guidance and information regarding which forms of physical activity are appropriate and which should be avoided. Patients who have had a myocardial infarction experience that physical activity evokes both hope and concern. It is important that the nursing care is grounded in a humanistic view of the person, acknowledging each patient as unique.

Background: Palliative home care has become increasingly common as more patientswish to be cared for and die at home. For family caregivers, this entails a transformationof everyday life, combining responsibility, emotional strain, and close involvement in thepatient’s care. District nurses play a key role in supporting both patients and familymembers throughout the palliative process. Aim: The aim of this literature review was toexplore aspects that influence the involvement of family caregivers’ in palliative homecare. Method: This study is an integrative literature review based on 15 qualitativeresearch articles published between 2015 and 2025. The articles were analyzed usingBettany-Saltikov and McSherry´s nine steps model. Results: The findings reveal thatfamily caregivers’ sense of participation is shaped through a dynamic interplay betweencommunication, relationships, support, and personal values. Open and empatheticdialogue with the district nurse fosters trust, meaning, and connection. Furthermore,information, emotional support, and affirmation of the caregiving role are identified askey conditions for participation. Cultural, ethical, and existential factors also influencehow caregivers perceive responsibility and meaning in care. Conclusion: The districtnurse’s caring presence, communicative competence, and relational continuity areessential for strengthening family caregivers’ participation and well-being in palliativehome care. The study highlights the need for further research to explore how districtnurses can develop sustainable strategies to support family caregivers’ emotional,existential, and practical needs. 

Studies indicate that it is well-known that leisure is important for adolescents’ health and improvement. Despite this, it´s not certain that all adolescents constantly have access to leisure activities constantly. That can depend on different reasons in the adolescent’s surroundings that prevent them from experiencing meaningful leisure. The reason for this study is to present what health-promoting measures are available to create meaningful leisure for adolescents in disadvantaged areas. The actors involved in young people’s leisure time are associations and recreational organizations. For this reason, the focus was placed on obtaining responses from individuals working within recreational activities who have a significant influence on young people. The method for this study was Qualitive, consisting of interviews. These interviews were conducted through personal meetings using a semi structured interview guide. The study shows leisure activities in disadvantaged areas are important for the adolescents because of the economic, cultural and social challenges they face. The youth coaches provide a safe and including environment where everybody can feel welcome. Young people are given the opportunity to develop and experience a sense of belonging by participating in fun and meaningful activities. They also get help through support from the recreational program and positive role models, take responsibility for one´s health and life situation, despite the challenges the young people may face. 

The following thesis aims to investigate how teachers and special educators in grade 1-3 attitudes are to work with differentiated teaching for students with dyslexia and reading- and writing difficulties in school. Using qualitative interviews with nine active teachers and special educators, their experinces and methods were analyzed. The result of the study shows that teachers´and special educators´experinces and access to resources affect their possibilities to offer their students a differentiated teaching. The result of the study showed that mapping the students´reading and writing skills early, switching working mode and close and good collaboration with a special educator were some of the most important elements for the teachers when finding methods to support students with dyslexia and reading- and writing difficulties. Challenges in their work were the lack of time and resources and big student groups. The result of the study indicates that relevant competence and school organizational conditions provide better conditions to promote students´knowledge to achieve goals in school. 

John Bean Technologies (JBT) is a global manufacturer of industrial freezing equipment, andtheir GC series consists of two-level spiral freezers used in large-scale food processing. Toaccess the upper level during service and cleaning, operators must climb a steep ladder andopen a hatch that functions as the entry point. This working posture creates an increased riskof falling, especially when operators carry tools or water hoses. JBT has therefore identifiedthe need for a safer and more ergonomic solution.The purpose of this bachelor thesis is to improve safety during climbing by developing a moreergonomic stair-ladder and by proposing a new opening mechanism for the hatch. The goalwas to create solutions compatible with all five sizes of the GC machine and to make dailywork easier for operators.The work was carried out in close collaboration with JBT and followed a structured processincluding a feasibility study, brainstorming, CAD modeling, and prototype development. Anew stair-ladder with a 65-degree angle was developed to reduce the risk of falling, and ahandrail was constructed and tested on the GC70 model. For the hatch, a concept using astainless-steel gas spring was proposed, suitable for cold and wet environments.Not all parts could be tested within the timeframe of the thesis, but JBT has received thecomplete material and plans to continue the development. The proposed improvements areexpected to increase safety, facilitate the work process, and have potential for implementationacross GC machines globally.

Background: Kidney failure is a disease that can occur suddenly or over a long period of time. The diagnosis usually entails lifelong dialysis treatment. There are two types of dialysis treatments, hemodialysis and peritoneal dialysis. Regardless of which dialysis treatment is used, the patient's autonomy and quality of life are affected. By understanding the patient's experience, the nurse can provide better patient centered care. Aim: The purpose was to highlight the patients´ experiences of living with dialysis and its impact on quality of life. Method: A general literature review with an inductive approach which was analyzed according to Popenoe et al. (2021). The results in the literature were conducted using nine scientific articles. Result: Based on the results, four main categories emerged: “Limitations in everyday life”, “Physical symptoms”, “Psychological consequences” and “Lack of information from healthcare professionals”. Conclusion: Dialysis patients experience major limitations in their everyday lives, which include both physical, psychological but also social obstacles. Loneliness is a common problem. Dialysis treatment may reduce patients’ work capacity, which often results in financial difficulties. Dialysis patients also experience great uncertainty about their diagnosis and dialysis treatment, due to a lack of information from healthcare professionals. Further research is needed on family and patient education about the disease and dialysis treatment for improved patient centered care. 

This research investigates students’ perceptions of ChatGPT as a feedback provider for programming assignments. Researcher interviewed undergraduate programming students and anchored our analysis in the Technology Acceptance Model (TAM). The result of analysis yielded six thematic insights into their experiences. Students often described ChatGPT as a helpful tutor; it provided clear, personalized explanations and exercises that boosted their understanding and confidence. Many students reported improved problem-solving and learning when using ChatGPT. However, students also expressed caution regarding ChatGPT’s limitations. They noted occasional errors in its feedback and stressed the need to verify suggestions. Overall, the findings of this research suggest that ChatGPT can effectively support programming education when used critically. Conclusion indicates that educators might integrate ChatGPT as a supplemental learning tool while guiding students to critically evaluate its feedback. This study provides a deeper understanding of AI-assisted feedback from the students’ perspective.