Publikationer från Uppsala Universitet
Not a member yet
127603 research outputs found
Sort by
Advancing CAR-based cell therapies for solid tumours : challenges, therapeutic strategies, and perspectives
No full textChimeric antigen receptor-cell therapies have demonstrated remarkable success in haematological malignancies but face significant hurdles in solid tumours. The hostile tumour microenvironment, antigen heterogeneity, limited tumour infiltration, and CAR-cell exhaustion contribute to reduced efficacy. Additionally, toxicity, off-target effects, and manufacturing challenges limit widespread clinical adoption. Overcoming these barriers requires a multifaceted approach that enhances CAR-cell persistence, trafficking, and tumour-specific targeting. Recent advancements in alternative cellular therapies, such as CAR-natural killer cells, CAR-macrophages, gamma delta CAR-T cells, and CAR-natural killer T cells, provide promising avenues for improving efficacy. These strategies leverage distinct immune cell properties to enhance tumour recognition and persistence. Furthermore, combination therapies, including chemotherapy, radiotherapy, antibodies, small molecule inhibitors, cancer vaccines, oncolytic viruses, and multi-CAR cell combination therapy, offer synergistic potential by modulating the TME and improving CAR-cell functionality. This review explores the challenges of CAR-based cellular therapies in solid tumours and highlights emerging strategies to overcome therapeutic limitations. By integrating novel cellular platforms and combination approaches, we seek to provide insights into optimising CAR-cell therapies for durable responses in solid malignancies
Charting circumstellar chemistry of carbon-rich asymptotic giant branch stars : II. Abundances and spatial distributions of CS
No full textContext: The circumstellar envelopes (CSEs) of asymptotic giant branch (AGB) stars harbour a rich variety of molecules and are sites of complex chemistry. Our current understanding of the circumstellar chemical processes of carbon-rich AGB stars is predominantly based on observations of a single star, IRC +10 216, often regarded as an archetypical carbon star. Aims: We aim to estimate stellar and circumstellar properties for five carbon stars, and constrain their circumstellar CS abundances. This study compares the CS abundances among the sources, informs circumstellar chemical models, and helps to assess if IRC+10 216 is a good representative of the physics and chemistry of carbon star CSEs. Methods: We modelled the spectral energy distributions (SEDs) and CO line emission to derive the stellar and outflow properties. Using these, we then retrieved CS abundance profiles with detailed radiative transfer modelling, imposing spatial and excitation constraints from ALMA and single-dish observations. Results: We obtain good fits to the SEDs and CO lines for all sources and reproduce the CS line emission across various transitions and apertures, yielding robust estimates of the CS abundance profiles. Peak CS fractional abundances range from 1x10-6-4x10-6, with e-folding radii of 1.8x1016-6.8x1016 cm. We also derive reliable 12C/13C and 32S/34S ratios from CS isotopologue modelling. Conclusions: Our results refine previous single-dish CS abundance estimates and improve the relative uncertainty on the CS e-folding radius for IRAS 07454-7112 by a factor of ∼ 2.5. Chemical models reproduce our estimates of the CS radial extent, corroborating the CS photodissociation framework used therein. We find no significant differences between the derived CS abundance profiles for IRC +10 216 and the rest of the sample, apart from the expected density-driven variations
Sales Prediction and Inventory Optimization for Vending Machines using XGBoost and Long Short-Term Memory
No full textAccurate forecasting of product demand in vending machines is crucial for effective inventory management, restocking efficiency, and minimizing waste. This thesis evaluates the performance of two machine learning models, eXtreme Gradient Boosting (XGBoost) and Long Short-Term Memory (LSTM), on the task of sales prediction for vending machines across Sweden. Using a dataset of over 5 million transactions from 2020 to 2023, the models predict weekly product sales in four-week windows throughout 2024. Evaluation is based on Mean Absolute Error (MAE), a custom Capped Mean Percentage Error (MPE), and MAE for active sales. XGBoost achieved the lowest overall error across all metrics, with the smallest standard deviation, and significantly shorter training times compared to LSTM, making it the preferred model for real-world deployment. The model was used to generate product recommendations based on predicted potential revenue (in SEK) and was evaluated in real-world conditions by swapping products in selected vending machines. While initial results indicate increased sales for several swapped products, a larger study is needed to validate the long-term impact
The First Man : En jämförelse av Adam som arketyp i Bibeln och i Hazbin Hotel
No full textThe character Adam in Amazon Primes animated hit series Hazbin Hotel is based on Adam from Genesis in the Bible. This essay examines whether the character from Genesis in the Bible and the character from the series Hazbin Hotel overlap in some way, and investigates whether the characteristics of Adam in Hazbin Hotels have any basis in the Bible. It also explores if Adam in Hazbin Hotels view on women can be traced back to Adam in Genesis view on women. Furthermore, it analyses Hazbin Hotels Adams relationship with the women around him in the story, and if he for example objectifies them, looks up to them or merely sees them as his property. This analysis is then compared to how Adam in the Bible treats Eve, and how he views her and his dominion over her after the Fall
Towards Model Merging for Tabular Telecommunications Data
No full textIn telecommunications, machine learning models are trained to optimize network functions such that connectivity can be leveraged to devices. With an increasing complexity in the network infrastructure, the number of machine learning models to manage also increases. Model merging aims to combine the capabilities of multiple machine learning models into a singular model, which could reduce the number of models needing management. With large, pre-trained models being made available, model merging has shown promise in language and computer vision settings using foundation models. However, current research has not investigated its applications using models trained on tabular data, despite it being the most common data modality. Most model merging methods base themselves on the linear mode connectivity, that is when the parameters of multiple machine learning are connected via a linear curve of near-constant loss. This study investigates the merging of multilayer-perceptrons trained on tabular telecommunications data. Specifically, we study the emergence of linear mode connectivity between two models trained on the same task via weight matching and via spawning, and compare a small selection of spawning-based merging methods when linear mode connectivity is attained. Results show that weight matching was insufficient on its own to make two models linear mode connected in multiple settings, while spawning did yield linear mode connected models in a majority of cases. The choice of merging methods when linear mode connectivity was attained did not show significant performance impact, but the utilization of a validation set to find the merging parameters appeared to be the strongest contributing factor. The results highlight that linear mode connectivity is not unique to text and image modalities, but requires further research on tabular data before it can be adopted into machine learning workflows in telecommunications
Food Security Through Food Dialogue? : Exploring Political Opinions on the Implementation of the EU Strategic Dialogue’s Policy Recommendations
No full textThe European food system contributes to environmental degradation and social discontent, promoting widespread protests by farmers and environmentalists across Europe. In response to criticism of the EU Green Deal and the Farm-to-Fork Strategy, the European Commission launched the EU Strategic Dialogue – a multi-stakeholder initiative aimed at shaping the future of the EU Food system. Through a quantitative content analysis, this thesis investigates the extent to which the 14 policy recommendations emerging from the Dialogue report address the six dimensions of food security, as defined by the High-Level Panel of Experts on Food Security and Nutrition. In addition, the study explores via a survey how political youth organizations in Sweden perceive the recommendations’ implementability. The findings reveal that the coverage of food security is uneven, even if the policy recommendations relate to all six food security dimensions. Most noticable, sustainability emerges as a cross-cutting theme. Moreover, most political youth organizations express broad support for the policy recommendations as adequate measures for food security. However, only a subset of the policy recommendations would likely gain majority support if the youth wings of parliamentary parties were to decide, highlighting political divisions despite shared concern for food security.
Den förbisedda faktorn : En kvantitativ studie om sambandet mellan kroppslängd och upplevd ensamhet i Europa
No full textKroppslängd är en fysisk egenskap som ofta förbises i forskning om sociala relationer och psykisk hälsa, trots dess dokumenterade betydelse för hur människor uppfattas och bemöts i samhället. Denna studie undersöker sambandet mellan kroppslängd och upplevd ensamhet bland vuxna i Europa, med särskilt fokus på könsskillnader och geografisk variation. Studien baseras på data från European Union Loneliness Survey (2024) och omfattar ett stratifierat urval av 3 436 respondenter från fyra europeiska länder med olika genomsnittlig kroppslängd: Nederländerna, Finland, Rumänien och Italien. Med en kvantitativ ansats genomförs enkla och linjära regressionsanalyser för att undersöka hur angiven kroppslängd relaterar till ett sammansatt index för upplevd ensamhet. Ensamhet mäts genom en sammansättning av tre etablerade instrument: De Jong Gierveld-skalan, UCLA Loneliness Scale samt ett direkt självskattat ensammhetsmått. Resultaten visar ett svagt men statistiskt signifikant negativt samband mellan kroppslängd och ensamhet bland kvinnor i Italien. Inga motsvarande samband återfanns bland kvinnor i övriga länder, eller bland män i något land. Resultaten tolkas i ljuset av Status Characteristics Theory (SCT), som belyser hur kroppslängd kan fungera som en statussignal och påverka sociala förväntningar, självpresentation och bemötande. Studien bidrar med ny kunskap genom att använda faktisk längdinformation i ett europeiskt vuxenurval, vilket särskiljer den från tidigare forskning som ofta utgått från ungdomar eller självupplevda uppfattningar om kroppslängd. Sammantaget visar studien att det inte finns något generellt eller substantiellt samband mellan kroppslängd och upplevd ensamhet bland vuxna i Europa. Ett svagt, statistiskt signifikant, negativt samband identifierades endast bland kvinnor i Italien, vilket tyder på att sambandet kan vara begränsat till specifika kulturella och könsspecifika kontexter. Detta fynd står i kontrast till tidigare forskning som huvudsakligen fokuserat på ungdomar och ofta baserats på subjektiva längduppfattningar. Därför bör kroppslängd betraktas som en möjlig, men kontextberoende, riskfaktor för ensamhet i vuxen ålder. Resultaten understryker behovet av fortsatt forskning om hur kroppslängd kan fungera antingen som en social resurs eller som en begränsning.
Kvalitetsdriven integration vid företagsförvärv inom vårdsektorn: Metoder och utmaningar : En studie om förbättringsmöjligheter i integrationsarbetet efter företagsförvärv
No full textIntegrationsprocesser efter företagsförvärv är komplexa och påverkar både struktur, kultur och arbetsmetoder. I denna studie undersöks hur verksamhetschefer och ledare inom förvärvade vårdenheter upplever integrationsarbetet efter förvärv samt vilka verktyg och stödformer som efterfrågas för att driva förändringen framgångsrikt. Studien syftar också till att identifiera hur en hållbar och inkluderande integrationsplan kan utformas.Arbetet har genomförts som en kvalitativ fallstudie på ett företag, i vårdsektorn som vuxit genom flera förvärv. Datainsamlingen har bestått av semistrukturerade intervjuer, en enkät samt dokumentanalys. Den teoretiska ramen bygger på kvalitetstekniska modeller som PDSAcykeln, förändringsledningsmodeller såsom Lewins trefasmodell och Kotters åttastegsmodell, samt teorier om organisationskultur.Resultaten visar att cheferna upplever ett starkt stöd i inledningen av integrationsprocessen, men att detta minskar över tid. Bristande kommunikation, otydlig ansvarsfördelning och avsaknad av långsiktig uppföljning leder till osäkerhet. Samtidigt efterfrågas verktyg för förankring, tydliggörande av roller och stöd i förändringsledarskap.Slutsatsen är att ett initialt stöd inte räcker. För att nå ett hållbart resultat krävs en systematisk integrationsstrategi där verksamhetsnära stöd, kulturförståelse och kontinuerlig förändringsledning är centrala komponenter
Euclid preparation LXIX. The impact of relativistic redshift-space distortions on two-point clustering statistics from the Euclid wide spectroscopic survey
No full textMeasurements of galaxy clustering are affected by redshift-space distortions (RSDs). Peculiar velocities, gravitational lensing, and other light-cone projection effects modify the observed redshifts, fluxes, and sky positions of distant light sources. We determined which of these effects leave a detectable imprint on several two-point clustering statistics to be extracted from the Euclid wide spectroscopic survey (EWSS) on large scales. We generated 140 mock galaxy catalogues with the survey geometry and selection function of the EWSS and made use of the LIGER (LIght cones with GEneral Relativity) method to account for a variable number of relativistic RSDs to linear order in the cosmological perturbations. We estimated different two-point clustering statistics from the mocks and used the likelihood-ratio test to calculate the statistical significance with which the EWSS could reject the null hypothesis that certain relativistic projection effects can be neglected in the theoretical models. We find that the combined effects of lensing magnification and convergence imprint characteristic signatures on several clustering observables. Their signal-to-noise ratio (S/N) ranges between 2.5 and 6 (depending on the adopted summary statistic) for the highest-redshift galaxies in the EWSS. The corresponding feature due to the peculiar velocity of the Sun is measured with a S/N of order one or two. The multipoles of the power spectrum from the catalogues that include all relativistic effects reject the null hypothesis that RSDs are only generated by the variation in the peculiar velocity along the line of sight with a significance of 2.9 standard deviations. As a by-product of our study, we demonstrate that the mixing-matrix formalism to model finite-volume effects in the multipole moments of the power spectrum can be robustly applied to surveys made of several disconnected patches. Our results indicate that relativistic RSDs, in particular the contribution from weak gravitational lensing, cannot be disregarded when modelling two-point clustering statistics extracted from the EWSS
Pooled optical screening in bacteria using chromosomally expressed barcodes
No full textOptical pooled screening is an important tool to study dynamic phenotypes for libraries of genetically engineered cells. However, the desired engineering often requires that the barcodes used for in situ genotyping are expressed from the chromosome. This has not previously been achieved in bacteria. Here we describe a method for in situ genotyping of libraries with genomic barcodes in Escherichia coli. The method is applied to measure the intracellular maturation time of 84 red fluorescent proteins