Publikationer från Umeå universitet
Not a member yet
55455 research outputs found
Sort by
A systematic review of intermediate fusion in multimodal deep learning for biomedical applications
No full textDeep learning has revolutionized biomedical research by providing sophisticated methods to handle complex, high-dimensional data. Multimodal deep learning (MDL) further enhances this capability by integrating diverse data types such as imaging, textual data, and genetic information, leading to more robust and accurate predictive models. In MDL, differently from early and late fusion methods, intermediate fusion stands out for its ability to effectively combine modality-specific features during the learning process. This systematic review comprehensively analyzes and formalizes current intermediate fusion methods in biomedical applications, highlighting their effectiveness in improving predictive performance and capturing complex inter-modal relationships. We investigate the techniques employed, the challenges faced, and potential future directions for advancing intermediate fusion methods. Additionally, we introduce a novel structured notation that standardizes intermediate fusion architectures, enhancing understanding and facilitating implementation across various domains. Our findings provide actionable insights and practical guidelines intended to support researchers, healthcare professionals, and the broader deep learning community in developing more sophisticated and insightful multimodal models. Through this review, we aim to provide a foundational framework for future research and practical applications in the dynamic field of MDL
Time-resolved hierarchical modeling highlights metabolites influencing productivity and cell death in Chinese hamster ovary cells
No full textBiopharmaceuticals are medical compounds derived from biological sources and are often manufactured by living cells, primarily Chinese hamster ovary (CHO) cells. CHO cells display variation among cell clones, leading to growth and productivity differences that influence the product's quantity and quality. The biological and environmental factors behind these differences are not fully understood. To identify metabolites with a consistent relationship to productivity or cell death over time, we analyzed the extracellular metabolome of 11 CHO clones with different growth and productivity characteristics over 14 days. However, in bioreactor processes, metabolic profiles and process variables are both strongly time-dependent, confounding the metabolite-process variable relationship. To address this, we customized an existing hierarchical approach for handling time dependency to highlight metabolites with a consistent correlation to a process variable over a selected timeframe. We benchmarked this new method against conventional orthogonal partial least squares (OPLS) models. Our hierarchical method highlighted several metabolites consistently related to productivity or cell death that the conventional method missed. These metabolites were biologically relevant; most were known already, but some that had not been reported in CHO literature before, such as 3-methoxytyrosine and succinyladenosine, had ties to cell death in studies with other cell types. The metabolites showed an inverse relationship with the response variables: those positively correlated with productivity were typically negatively correlated with the death rate, or vice versa. For both productivity and cell death, the citrate cycle and adjacent pathways (pyruvate, glyoxylate, pantothenate) were among the most important. In summary, we have proposed a new method to analyze time-dependent omics data in bioprocess production. This approach allowed us to identify metabolites tied to cell death and productivity that were not detected with traditional models
Geochemical transformations of gypsum under multiple environmental settings and implications for Ca-sulfate detection on Mars
No full textCalcium sulfate minerals are found in multiple environments on Earth and Mars, with chloride (Cl) salts widely distributed on both planets. Low-temperature studies have explored geochemical processes, including the formation of transient liquid water and ion migration on Mars. Some Cl-salts (e.g., NaCl and CaCl2) can dissolve gypsum (CaSO4·2H2O) in certain environments, making gypsum-Cl salt interactions significant. Additionally, gypsum’s geochemical transformation at high temperatures reveals dehydration pathways crucial for understanding Mars’ aqueous history and potential for life. This study examines gypsum dehydration through (i) thermal analyses and (ii) interactions with Cl-salts over a temperature range of −90 to 400 °C. We applied three spectroscopic techniques (Raman, visible/near-infrared, and mid-IR) plus X-ray diffraction (XRD) to analyze these samples under variable conditions. This study also provides a low-temperature spectral data set for gypsum and gypsum-Cl salt mixtures, beneficial for orbital analyses. Our findings reveal that experimental (i) heating rates, (ii) temperature ranges, (iii) relative masses of gypsum and Cl-salts, and (iv) dehydration environments (e.g., in situ and in vacuo) influence Ca-sulfate phase formation. Although we find different results in some cases, this study demonstrates that changing experimental conditions affects the detectability and transformation of gypsum. Further, these results indicate that the geochemical environmental conditions on Mars play a role in gypsum’s geochemical transformation to dehydrated components. This study also provides structural and chemical data for Ca sulfate assemblages from vibrational spectroscopy and XRD, which extends our knowledge of gypsum and related materials under variable conditions, thus aiding orbital and surface planetary analyses that may help to advance our understanding of planetary geochemistry on Mars
Social justice in HCI : current streams, considerations, and ways forward
No full textThe expanding interest in justice-oriented HCI focusing on critical perspectives, structural oppression, and marginalization—often referred to as social justice—is reflected in a growing number of publications over the past few years. Through the continuous growth of social justice in HCI, we argue that now is a good time to provide an overview of the ongoing and current streams of social justice research. We introduce social justice as it has grown in HCI during the last 15 years followed by the most commonly framed theoretical tenets. Secondly, we construct a corpus of 60 HCI articles building on social justice as the main concept. Through our corpus we summarize and present 4 currently ongoing streams of research to further a cohesive, yet fluid understanding of how social justice is shaped and understood within HCI. Describing these interconnected streams also gives us the possibility to frame and describe the current development and move forward as a research community. Based on our study and discussions, we suggest 6 considerations for HCI researchers seeking to work with social justice as a concept, and we emphasize the need for long-term engagement in justiceoriented research to foreground and enable societal change and social good. Through this study, we contribute to the ongoing growth of social justice research in HCI by providing an overview of current streams and ways forward
Improving locally differentially private graph statistics through sparseness-preserving noise-graph addition
No full textDifferential privacy allows to publish graph statistics in a way that protects individual privacy while stillallowing meaningful insights to be derived from the data. The centralized privacy model of differential privacyassumes that there is a trusted data curator, while the local model does not require such a trusted authority.Local differential privacy is commonly achieved through randomized response (RR) mechanisms. This doesnot preserve the sparseness of the graphs. As most of the real-world graphs are sparse and have several nodes,this is a drawback of RR-based mechanisms, in terms of computational efficiency and accuracy. We thus,propose a comparative analysis through experimental analysis and discussion, to compute statistics with localdifferential privacy, where, it is shown that preserving the sparseness of the original graphs is the key factorto gain that balance between utility and privacy. We perform several experiments to test the utility of theprotected graphs in terms of several sub-graph counting i.e. triangle, and star counting and other statistics. Weshow that the sparseness preserving algorithm gives comparable or better results in comparison to the otherstate of the art methods and improves computational efficiency.57001135
Antibacterial compounds against non-growing and intracellular bacteria
No full textSlow- and non-growing bacterial populations, along with intracellular pathogens, often evade standard antibacterial treatments and are linked to persistent and recurrent infections. This necessitates the development of therapies specifically targeting nonproliferating bacteria. To identify compounds active against non-growing uropathogenic Escherichia coli (UPEC) we performed a drug-repurposing screen of 6454 approved drugs and drug candidates. Using dilution-regrowth assays, we identified 39 compounds that either kill non-growing UPEC or delay its regrowth post-treatment. The hits include fluoroquinolones, macrolides, rifamycins, biguanide disinfectants, a pleuromutilin, and anti-cancer agents. Twenty-nine of the hits have not previously been recognized as active against non-growing bacteria. The hits were further tested against non-growing Pseudomonas aeruginosa and Staphylococcus aureus. Ten compounds - solithromycin, rifabutin, mitomycin C, and seven fluoroquinolones-have strong bactericidal activity against non-growing P. aeruginosa, killing >4 log10 of bacteria at 2.5 µM. Solithromycin, valnemulin, evofosfamide, and satraplatin are unique in their ability to selectively target non-growing bacteria, exhibiting poor efficacy against growing bacteria. Finally, 31 hit compounds inhibit the growth of intracellular Shigella flexneri in a human enterocyte infection model, indicating their ability to permeate the cytoplasm of host cells. The identified compounds hold potential for treating persistent infections, warranting further comparative studies with current standard-of-care antibiotics.
This isn't journalism, it's propaganda! : patterns of news media bias accusations on Twitter, 2010–2020
No full textIn several countries, research shows that mainstream news media have increasingly become a target of political attacks, and a recurring accusation is that the news media are biased politically. Since social media play an important role in circulating accusations of media bias, in this paper we analyze 167,527 Swedish tweets about media bias over a ten-year period using the mixed-methods CADS approach, which incorporates computational processes with discourse analysis. Specifically, we explore the prevalence of bias accusations over time and which news outlets constitute the main targets; from what political perspectives these bias accusations have originated over time; and how discourses accusing the news media of political bias are constructed. Among other things, the results show that accusations of left-wing bias are much more common than accusations of right-wing bias. A key reason is that actors on the political right engage much more frequently in accusations of media bias than their center-left counterparts, and that discourses of media bias are heavily dominated by right-wing populist tropes in which mainstream media are part of the "corrupt", lying elite who spread "politically correct" disinformation
The promise of neuromorphic edge AI for rural environmental monitoring
No full textEdge AI is the fusion of edge computing and artificial intelligence (AI). It promises responsiveness, privacy preservation, and fault tolerance by moving parts of the AI workflow from centralized cloud data centers to geographically dispersed edge servers, which are located at the source of the data. The scale of edge AI can vary from simple data preprocessing tasks to the whole machine learning stack. However, most edge AI implementations so far are limited to urban areas, where the infrastructure is highly dependable. This work instead focuses on a class of applications involved in environmental monitoring in remote, rural areas such as forests and rivers. Such applications have additional challenges, including failure proneness and access to the electricity grid and communication networks. We propose neuromorphic computing as a promising solution to the energy, communication, and computation constraints in such scenarios and identify directions for future research in neuromorphic edge AI for rural environmental monitoring. Proposed directions are distributed model synchronization, edge-only learning, aerial networks, spiking neural networks, and sensor integration
ATT ORKA LEDA ÖVER TID : En studie om arbetsmiljö och hållbarhet för elittränare inom svensk ishockey
No full textStudien undersökte elittränarrollen inom svensk ishockey, med fokus på de faktorer som påverkar möjligheterna till ett hållbart ledarskap. Syftet var att fördjupa förståelsen för tränarnas upplevelser av arbetsbelastning, yttre press och det stöd och de förutsättningar som erbjuds inom deras klubbar, samt hur dessa faktorer påverkar tränarnas välmående och prestation. Studien utgick från tidigare forskning om elittränaryrkets komplexitet, krav-kontroll-(stöd)modellen och teorier om hållbart ledarskap. Datainsamlingen baserades på semistrukturerade intervjuer med åtta tränare verksamma i SHL och SDHL, och insamlad data analyserades med tematisk analys. Resultaten visade att tränarna verkade i en arbetsmiljö präglad av hög arbetsbelastning och ständig yttre press från media, sociala medier och klubbledning. Den yttre pressen påverkade främst tränarnas välmående, men kunde även ha en indirekt negativ effekt på prestation genom stress, sömnsvårigheter och begränsad återhämtning. Tillgången till stöd varierade kraftigt mellan klubbarna; i de flesta fall vilade ansvaret för att söka stöd på tränarna själva. Endast en klubb hade en etablerad intern stödfunktion. Sammantaget pekade resultaten på behovet av tydligare strukturer och proaktiva stödinsatser på organisationsnivå – såsom en etablerad HR-funktion – för att främja tränarnas långsiktiga hälsa och ett hållbart ledarskap. Studien rekommenderar ett systematiskt arbete med arbetsmiljöfrågor för att förebygga ohälsa och stärka tränarnas förutsättningar att orka leda över tid
From problem-solving, innovation and creativity to empathy, connection and care? : Troubling the use of STEAM buzzwords in early childhood education research
No full textThere is a growing trend of addressing the benefits of STEAM (science, technology, engineering, arts, and mathematics) in literature on Early Childhood Education (ECE). The literature often assumes that adding Arts to STEM in ECE will help young children develop a number of skills such as critical thinking, innovation, creativity, problem-solving, communication and collaboration. We refer to these skills as STEAM buzzwords since they are listed in a recurrent way throughout the literature and are seldom critically assessed or challenged. With this colloquium, we aspire to challenge the use of these buzzwords. The main reason is that three of them, innovation, creativity and problem-solving, carry a gendered and unjust history, associated with white men, progress, economic growth and conquest. We argue that an unreflective use of these buzzwords may steer STEAM education in ECE towards fostering ‘human capital’ rather than enabling children to develop close and empathic relations with organisms and other more than human actors and elements in their surrounding world. Therefore, we invite practitioners and researchers to join us in forming a new set of STEAM buzzwords, a set that is just and apt for all children, and for the world