Mid Sweden University
Not a member yet
15329 research outputs found
Sort by
Optimized Classification of Disturbed and Undisturbed Snow in Images for Energy-Efficient IoT Implementation
No full textThis project explores how to use lightweight deep learning models to distinguish between disturbed and undisturbed snow in images a task relevant for things like trail monitoring or outdoor safety in snowy areas. The idea is to make the model small and efficient enough to run on low-power IoT devices without needing the cloud. By collecting a custom image dataset and trained several modern models EfficientNetB0, Vision Transformer (ViT), and Swin Transformer using multiple image sizes. After testing them for accuracy and speed, I applied pruning and quantization to shrink the models and make them run faster. The best performing model was deployed using Edge Impulse to check real-world energy usage and inference time. Results showed that transformers were more accurate overall, but EfficientNet-B0 was better suited for energy limited devices, especially at a 256 × 256 resolution. This work shows a complete process from training to IoT deployment for image classification under real constraints, and provides insight into how different optimization techniques affect both performance and efficiency
Broken Access Control Risks in Open-SourceJavaScript Project: A Security Analysis
No full textContext: Broken Access Control (BAC) has beenidentified as the most critical web security issue by OWASP.Open-source JavaScript projects, while enablingcollaboration and transparency, remain particularlyvulnerable to BAC due to their visibility and diversecontributor base.Objective: This study investigates the prevalence andpatterns of BAC vulnerabilities in open-source JavaScriptprojects. The goal is to understand how these flaws manifest,how they can be systematically detected, and whethercurrent detection methods are effective in real-worldscenarios.Method: A hybrid methodology was applied consisting ofautomated static code analysis using Semgrep and manualpenetration testing. A GitHub mining script was used tocollect a curated dataset of 166 actively maintainedJavaScript repositories, selected based on access-controlrelated keywords and popularity metrics. Custom Semgreprules were developed to detect BAC categories includingInsecure Direct Object References (IDOR), unprotectedroutes, forced browsing, and token or sessionmisconfigurations. Manual validation was conducted inisolated Docker environments and Postman program toconfirm exploitability.Results: Of the 166 analyzed repositories, 33 were flagged bySemgrep, and 5 were confirmed to contain real, exploitableBAC vulnerabilities through manual testing. Validatedvulnerabilities included unauthenticated endpoints,parameter-based access to privileged functions, and insecureCORS or token handling. Although static analysis proveduseful in surfacing suspicious patterns, it exhibited a highfalse positive rate and lacked contextual understanding ofruntime authorization.Conclusion: The findings reveal recurring patterns of BACvulnerabilities in open-source JavaScript applications. Whilestatic tools like Semgrep proved useful for initial screening,they must be paired with manual validation to ensureaccurate assessment
From Lignocellulosic Residues to Protein Sources : Insights into Biomass Pre-Treatments and Conversion
No full textWith the global population steadily rising, the demand for sustainable protein sources has become increasingly urgent. Traditional animal- and plant-based proteins face challenges related to scalability, resource efficiency, and environmental impact. In this context, single-cell protein has emerged as a promising alternative. Derived from microorganisms such as algae, bacteria, fungi, and yeast, single-cell protein offers a high nutritional profile- including all essential amino acids and vitamins-while enabling rapid production, minimal land and water requirements, and no generation of greenhouse gas emissions. A particularly compelling advantage of single-cell protein is its ability to be produced from agro-industrial waste, converting low-cost residues into valuable nutritional resources and contributing to environmental sustainability. Among these waste streams, lignocellulosic biomass from agricultural and forestry residues stands out as a renewable, biodegradable, and abundant feedstock. This review explores the potential of lignocellulosic waste as a substrate for single-cell protein production, emphasizing both its environmental advantages and nutritional value. It highlights the single-cell protein role as a sustainable and scalable alternative to conventional protein sources. The review also identifies key scientific, economic, and regulatory challenges, and recognizes the importance of targeted investments, particularly in policy development, public awareness, and technological innovation, to enable the broader adoption and acceptance of single-cell protein -based products
Continental Differences in Industrial Water Usage and Environmental Policy Adaption : Insights from UN Global Compact
No full textThis report investigates global disparities in industrial water use and the implementation of related environmental policies, using quantitative data regarding industries sustainability reporting from UN Global Compact. The study analyzes industrial engagement with water policies in relation to other environmental topics, compares water consumption and withdrawal across continents and explores regional differences in regulatory adaptation, technological adoption and economic incentives. Findings show that water is among the least prioritized policy areas; trailing behind energy, waste and climate change. 53% of companies report water-specific policies, with Asia showing the highest adoption rate (60%) and Europe and Africa the lowest (49%). Data also reveals major variations in water consumption and withdrawal per continent, with Africa and Europe reporting the highest consumption and Latin America the highest withdrawals. These trends align with socioeconomic and industrial structures, as well as regulatory strength. The study highlights that voluntary frameworks can enhance transparency and prepare industries for future legislation, but impactful water stewardship requires binding policies, economic incentives and ecotechnological solutions. These insights contribute to a better understanding of global industrial water governance and inform strategies for sustainable resource management
Pensionärer och livslångt lärande : En kvalitativ intervjustudie
No full textSyftet med denna studie var att beskriva och analysera pensionärers upplevelser av lärande efter arbetslivets slut. Studien utgick från tre forskningsfrågor där den första handlade om på vilka sätt pensionärer uppger att de engagerar sig i lärande efter arbetslivets slut. Den andra forskningsfrågan berörde vad pensionärer beskriver att de lär sig efter arbetslivets slut medan den tredje forskningsfrågan undersökte vad pensionärer upplever som meningsfullt att lära sig efter arbetslivets slut. Det sociokulturella perspektivet har varit det teoretiska ramverket. Det empiriska materialet samlades in genom kvalitativa intervjuer med tio pensionärer. Materialet analyserades med hjälp av kvalitativ innehållsanalys. Resultatet visar att pensionärer efter arbetslivets slut medvetet letar efter lärande vilket bidrar till ett aktivt åldrande. Digitala resurser möjliggör självstyrt lärande för pensionärer medan föreningsliv och kurser utgör sociala och situerade lärmiljöer. Pensionärer lär sig främst om samhällsengagemang, hushållssysslor samt hantverk och musik. Lärandet beskrivs som meningsfullt när det är vardagsnära, självvalt och kopplat till personlig utveckling. Slutsatserna är att pensionärer aktivt söker lärande i många olika former och sammanhang efter arbetslivets slut där språket blir betydelsefullt för lärande. Resultatet belyser vikten av att erbjuda aktiviteter där lärande kan möjliggöras för pensionärer.Godkänd 2025-05-23</p
Nil Per Os In The Emergency Department : A Quantitative Study On Fasting Before Presumptive Emergency Surgery
No full textIntroduction: Nil per os is a common precautionary practice in emergency departments for patients with suspected need for emergency surgery. However, this practice may lead to prolonged fasting periods, which can negatively affect patient outcomes. This study aimed to evaluate the necessity and appropriateness of nil per os for patients in the emergency department. Methods: A retrospective quantitative observational study was conducted, analyzing 41,983 patient visits from 2 emergency departments in 2023. The study focused on surgical, orthopedic, and neurological chief complaints, where nil per os was commonly implemented. We assessed the negative predictive value of emergency surgery within specified time intervals (<6 hours and <12 hours after arrival) solely based on time from arrival in combination with surgical priorities, chief complaints, and triage priorities according to the Rapid Emergency Triage and Treatment System. Results: Among the 41,983 included patients, 6.1% (n = 2561) underwent emergency surgery. The negative predictive value for emergency surgery within the specified time intervals was highest for Rapid Emergency Triage and Treatment System priorities 3, 4, and 5 (>99%), indicating a very low likelihood of emergency surgery within 6 to 12 hours for these patients. Rapid Emergency Triage and Treatment System priority 1 had the lowest negative predictive value, approximately 80% to 90%. Discussion: The study indicates that the routine practice of nil per os upon arrival at the emergency department is often unwarranted, particularly for those not classified as Rapid Emergency Triage and Treatment System priority 1. These findings highlight the need for updated guidelines to ensure that nil per os protocols are issued based on a clear medical necessity, thereby minimizing unnecessary fasting and its associated risks
Tillsynens verkan och relevans : Sju fallstudier om tillsyn över kommunala myndigheters arkiv i Sverige
No full textDen här studien ämnar undersöka tillsynens påverkan och relevans för kommunala myndigheters arkivverksamhet. För att uppnå syftet har frågeställningar ställts till deltagare vid inspektioner som rör synpunkter och upplevelser av en utförd tillsyn samt tillsynens påverkan. Studien bygger på en kvalitativ metod och omfattar sju fallstudier med fokus på kommunala myndigheter som genomgått en inspektion utförd av kommunernas arkivmyndigheter. Den huvudsakliga datainsamlingstekniken utgjordes av semistrukturerade intervjuer tillsammans med dokumentation som skapas vid tillsyn samt en litteraturgenomgång över relaterad forskning. Det teoretiska ramverket som används för att analysera resultatet är implementeringsteori samt records continuum-modellen. Studiens resultat visar att deltagarna vid inspektionerna upplever arbetet innan och efter en inspektion som tidskrävande, men ändock viktigt. Resultatet visar också ett perspektiv där brist på kunskap och resurser utgör ett hinder både före och efter inspektionerna. Upplevelserna av arbetet kopplat till tillsyn skiljer sig åt beroende på vilka resurser kommunerna har att arbeta med. Tillsynens relevans påverkas också av detta. Det framkommer vidare att tillsynernas utförande över lag upplevdes som bra. Det finns dock synpunkter rörande tillsynernas utförande som kan ha haft effekter på tillsynernas påverkan och relevans. De fysiska arkiven fick mindre fokus i de fall där inspektionen utfördes digitalt. Den digitala inspektionen medför också att arkivmyndigheten inte får en verklig bild av den inspekterade myndighetens situation. Tillsynen kan även upplevas som okritisk. Tillsynens utförande upplevs också som lärande och stödjande och har därför blivit ett viktigt verktyg för arkiv- och informationshanteringen. Tillsynen har i flera fall påverkat arkivverksamheterna på så sätt att de påpekande bristerna åtgärdats. Det gäller för de myndigheter som haft rätt förutsättningar i form av kunskap, vilja och resurser. Tillsynen har också påverkat uppmärksamheten och medvetenheten för arkivfrågor inom myndigheterna. Det gäller för alla myndigheter. This study investigated the impact and relevance of archival audit inspections for municipal authorities' archive activities. To achieve the aim and to answer the research questions, an interview guide with semi structured questions was used to access the views and experiences of inspection participants on the impact of the inspections. The study was based on a qualitative methodology and included seven case studies focusing on municipal authorities that have undergone an inspection, carried out by the municipal archive authorities. The main data collection technique consisted of semistructured interviews together with documentation created during inspection and a literature review of related research. The theoretical framework used to analyze the results was Implementation Theory and the Records Continuum Model. The results of the study show that some experienced the work before and after an inspection as time-consuming, but still important. The results also revealed a perspective where lack of knowledge and resources constituted an obstacle both before and after the inspections. The experiences of the inspection participants on inspections differed depending on the resources the municipalities had at hand. This also meant that inspections impacted the municipal authorities differently. The relevance of inspections was also affected by this. The findings show that the inspections are perceived as good. However, there were varying views regarding the way the inspections were performed and the impact they had on the archival activities. The physical archives received less attention where the inspection was digitally carried out. In such a case, it was revealed that the archival authority did not get a holistic picture of the inspected authority. The audit inspection could also be perceived as uncritical. They were, however, also perceived by some as educative and hence a supportive tool towards the management of archives. In several cases, the audit inspections had a positive impact on the archival operations and hence the correction of the deficiencies. This applied to the municipal authorities that had the knowledge, willingness and human resources. Furthermore, the inspections raised awareness and increased the attention paid to archival issues within the authorities. This was the case with all the authorities.
‘Still struggling, but there is a small glimmer of light at the end of the tunnel’ – Individual in-depth interviews six-months after stillbirth : A follow-up study in Sweden
No full textBackground: Few parents experience stillbirth in Sweden, and their perspectives on the grieving process remain largely unknown. Objective: To explore parents’ perspectives, memories, reflections and insights in the grieving and recovery process six months after stillbirth. Methods: A mixed-method study involving nine in-depth interviews and responses to eleven quantitative statements. Data were analysed using descriptive statistics and thematic network analysis. Results: Two themes were identified ‘Retained indelible deep memories of the loss and increased insights into the tragic life-changing event’ and ‘The continued grieving process, quality of life and new pregnancy’. These themes captured parents’ ongoing grief and reflections. Six months later, parents reported gaining insights into unforeseen pregnancy abnormalities. Some announced new pregnancies, desiring extended medical checks by both midwives and obstetricians. There was a widespread request for tailored psychological support throughout pregnancy. Qualitative findings were reinforced by the quantitative statements. Conclusions: Stillbirth profoundly affects parents long-term, with deep grief and memories. They develop coping strategies to strengthen their mental health. However, after six months, they begin to see a small glimmer of light at the end of the tunnel.
Artificiell intelligens, tillgänglighet och handlingsoffentlighet : En kvalitativ studie av Riksarkivets användning av AI
No full textSyftet med denna studie var att bidra till kunskapen kring vilken inverkan av användningen av artificiell intelligens har på tillgängligheten till arkiven, samt undersöka hur lagstiftning, i synnerhet rörande handlingsoffentlighet, påverkar AI-användningen. För detta genomfördes en kvalitativ fallstudie av Riksarkivet och dess användning av AI. Insamlad data bestod huvudsakligen av dokumentation som Riksarkivet själva gjort tillgängligt på deras webbplats. Datan analyserades med bakgrund av relaterad forskning samt huvuddragen i relevant lagstiftning. Av analysen av datan drogs slutsatsen att AI i huvudsak har positiv inverkan på tillgängligheten. Dess användning möjliggör att användaren på egen hand lättare kan eftersöka och ta del av handlingarna, det öppnar upp möjligheter till nya forskningsfrågor och -områden som inte tidigare var möjliga, och det bidrar även till en effektivisering av handläggning av förfrågningar på de samlingar som inte kan tillgängliggöras i sin helhet. Den enda möjliga nackdelen för tillgängligheten som kunnat identifieras är frågan om urval av vilka samlingar som ska tillgängliggöras med AI, men som det framgår av det studerade fallet så kan detta hanteras dels genom att föra en dialog med forskningssamhället, dels genom att se till den direkta nyttan av AI-användningen med en viss samling. I det studerade fallet syns inga uttalade begränsningar som kommer av offentlighets- och relaterad lagstiftning. De begränsningar kring hantering av personuppgifter och sekretessbelagd information som tidigare forskning belyst tycks i dagsläget hanteras genom att använda AI på samlingar där lagstiftning såsom OSL och GDPR inte är aktuella, respektive användning på alternativa sätt på de samlingar som skulle kunna omfattas av lagstiftningen. En möjlig positiv effekt av offentlighetsprincipen kan synas i det att man använder sig av crowdsourcing i arbetet med AI, men i övrigt har inga konkreta positiva effekter av handlingsoffentligheten kunnat härledas. Vidare studier krävs.This study aimed to contribute to the knowledge base regarding artificial intelligence and its impact on accessibility of archives, as well as to examine how legislation related to freedom of information affects AI usage. This was accomplished by conducting a qualitative case study of the Swedish National Archives (Riksarkivet) and its use of AI. Collected data mainly consisted of documentation made available by Riksarkivet on their website. Data analysis was conducted based on existing research and the main points of relevant laws. The conclusion reached was that AI mainly has a positive impact on the accessibility of archives. It facilitates a higher degree of user independence in searching and access, it enables access to new research domains previously unexplored, and it contributes to increased efficiency in managing access requests for collections that are not suitable for publication in their entirety. The only possible drawback for accessibility that was able to be identified was the matter of evaluation of what collections on which to implement AI. However, as was apparent in the case in question, this can be handled by consulting the research community, as well as by considering the direct benefit of increased accessibility of a given collection. In the case at hand, no explicit limitations stemming from freedom of information and such laws could be identified. Existing research outlines challenges in the management of personal and/or sensitive information. This is, in this case, circumvented by mainly using AI with collections where such laws are not applicable, or using it in alternate ways with collections where they are. A possible positive effect of the freedom of information might be apparent in Riksarkivet's usage of crowdsourcing. Other than that, however, no definite positive effects could be identified. Further research is needed
Kunskapsluckor i matematik på lågstadiet : Kartläggning, insatser och evidensbaserade undervisningsmetoder.
No full textLitteraturstudien analyserar och undersöker insatser för att förebygga kunskapsluckor i matematik. Studien fokuserar på hur insatserna kan utformar, vilka insikter som behövs för att utforma dem samt hur lärare kan tillägna sig dessa insikter. Forskningen i litteraturstudien visar att tidiga insatser spelar en avgörande roll för att förebygga risken för kunskapsluckor i matematik. I litteraturstudien genomförs en systematisk tematisk analys utifrån förutbestämda teoretiska perspektiv. Litteraturstudien analyseras forskning om kartläggning, kognitiva processer, socioemotionella faktorer och evidensbaserade undervisningsmetoder. Datainsamlingen genomfördes i DIVA, Google Scholar och MIUNS bibliotek, där peer-reviewed artiklar med fokus på matematikundervisning och insatser identifierades. I Urvalsprocessen säkerställdes att enbart empiriska studier med relevanta resultat inkluderades. Resultaten visar att systematiska kartläggningar är viktiga för att identifiera elevernas kunskaper och i sin tur kunna skapa individanpassade insatser som baseras på explicita undervisningsstrategier. Exekutiva funktioner visar sig ha en central roll i utvecklingen av insatserna. Studien bidrar med en översikt av effektiva pedagogiska strategier och ger praktiska rekommendationer för hur lärare kan utveckla undervisningen för att stödja elevers långsiktiga matematiska kunskaper.Godkänd 2025-05-23</p