| 51. |
- Johansson, Peter, Associate professor, 1978-, et al.
(författare)
-
Disrupting the research process
- 2020
-
Ingår i: LUT Scientific and Expertise Publications. - 9789523354661
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- As the research process, i.e. how researchers conduct research, may be disrupted by Artificial Intelligence (AI)-based tools, this research in progress stresses the need for research on the future role of researchers. Prior research on AI-based tools used in the research process, focuses primarily on technological development related to, for example, language recognition, text extraction, and paper summaries. Recently, scientific texts are generated by AI. In this research in progress, four sub-areas of interest are identified: Black-box dilemma. What part or parts of the research process, if any, can trustfully be replaced by AI-based tools? Quality. What is required for AI-based tools to become an integrated part of the research process? Benefits/limitations. What are the potential benefits and limitations with AI-based tools as an integrated part of the research process? Progress in use: What is the progress of the use and development of AI-based tools for research? Research design, potential contribution, and practical implications of this research are discussed.
|
|
| 52. |
- Johansson, Torbjörn, et al.
(författare)
-
Datadrivna metoder för att detektera avvikande mätvärden inom dammsäkerhet
- 2020
-
Rapport (övrigt vetenskapligt/konstnärligt)abstract
- Vi gör en litteraturstudie av state of the art i dammsäkerhet och tar reda på vad som har gjorts i andra länder. Vi tittar även på teorier och metoder som inte ännu har använts i dammsäkerhet. Det är lätt att komma igång med datadriven analys med hjälp av färdiga verktygslådor som finns öppet tillgängliga. Vi diskuterar därför kring vanliga fallgropar vid dataanalys och hur man undviker dem. Mycket av den vetenskapliga litteraturen inom datadriven dammsäkerhet handlar om tillämpning av maskininlärningsmetoder. Den struntar ofta i förbehandling och antar att alla sensorer fungerar. Vi summerar resultat om förbehandling av sensordata och hur avvikelser kan detekteras, och ser att här finns ett gap i dammsäkerhetslitteraturen. Vi ser att väl förstådda och utredda metoder som PCA och PLS, tidsseriemodellering, SPC/MSPC och neurala nätverk är lämpliga för vidare studier. Med stöd av tiotals års erfarenhet av forskning och utveckling i datadrivna metoder ger vi rekommendationer för hur ett arbete för att implementera datadrivna metoder för att höja dammsäkerheten i Sverige skulle kunna utformas. Vi ser att datadrivna metoder för dammsäkerhet är redo att implementeras i Sverige, och bedömer att i kombination med väl utformad instrumentering och rätt givarplacering kan de bidra till väsentligt höjd dammsäkerhet.
|
|
| 53. |
- Johansson, Viktoria, et al.
(författare)
-
Twin study shows association between monocyte chemoattractant protein-1 and kynurenic acid in cerebrospinal fluid
- 2020
-
Ingår i: European Archives of Psychiatry and Clinical Neuroscience. - : Springer Science and Business Media LLC. - 0940-1334 .- 1433-8491. ; 270:7, s. 933-938
-
Tidskriftsartikel (refereegranskat)abstract
- Preclinical studies indicate a link between the kynurenine pathway and monocyte chemoattractant protein-1 (MCP-1), but there is a lack of clinical studies examining this further. We here perform a secondary analysis of kynurenine metabolites and MCP-1 in cerebrospinal fluid of 23 twins affected from schizophrenia, bipolar disorder or unaffected. We show an association between MCP-1 and kynurenic acid (KYNA), driven by unique environmental influences and a less pronounced association between MCP-1 and tryptophan. No association was detected between MCP-1 and quinolinic acid. Further studies on the mechanism behind the putative relationship between KYNA and MCP-1 are needed. © 2019, The Author(s).
|
|
| 54. |
- Johansson Wensman, Jonas, et al.
(författare)
-
Application of Viral Metagenomics for Study of Emerging and Reemerging Tick-Borne Viruses
- 2020
-
Ingår i: Vector-Borne and Zoonotic Diseases. - : Mary Ann Liebert Inc. - 1530-3667 .- 1557-7759. ; 20, s. 557-565
-
Tidskriftsartikel (refereegranskat)abstract
- Ticks are important vectors for different tick-borne viruses, some of which cause diseases and death in humans, livestock, and wild animals. Tick-borne encephalitis virus, Crimean-Congo hemorrhagic fever virus, Kyasanur forest disease virus, severe fever with thrombocytopenia syndrome virus, Heartland virus, African swine fever virus, Nairobi sheep disease virus, and Louping ill virus are just a few examples of important tick-borne viruses. The majority of tick-borne viruses have RNA genomes that routinely undergo rapid genetic modifications such as point mutations during their replication. These genomic changes can influence the spread of viruses to new habitats and hosts and lead to the emergence of novel viruses that can pose a threat to public health. Therefore, investigation of the viruses circulating in ticks is important to understand their diversity, host and vector range, and evolutionary history, as well as to predict new emerging pathogens. The choice of detection method is important, as most methods detect only those viruses that have been previously well described. On the other hand, viral metagenomics is a useful tool to simultaneously identify all the viruses present in a sample, including novel variants of already known viruses or completely new viruses. This review describes tick-borne viruses, their historical background of emergence, and their reemergence in nature, and the use of viral metagenomics for viral discovery and studies of viral evolution.
|
|
| 55. |
- Karagianni, A., et al.
(författare)
-
Recurrent cerebrovascular events in patients after percutaneous closure of patent foramen ovale
- 2020
-
Ingår i: Journal of Stroke & Cerebrovascular Diseases. - : Elsevier BV. - 1052-3057. ; 29:8
-
Tidskriftsartikel (refereegranskat)abstract
- Background: Recent published trials have shown the efficacy of transcatheter closure of patent foramen ovale (PFO) for preventing recurrent cryptogenic cerebrovascular events. However, the risk of recurrent cerebrovascular events (rCVEs) is up to 5.7%, and the etiology is unclear. Objective: This study aimed to investigate the risk factors for rCVEs after closure of PFO during long-term follow-up. Methods: In our center, 282 consecutive patients underwent PFO closure because of a cryptogenic cerebrovascular event between 2006 and 2014. Their Risk of Paradoxical Embolism (RoPE) score was calculated retrospectively. We followed up with the patients by telephone, using hospital records to identify those who suffered from rCVEs. Patients with rCVEs were matched with two control patients of the same sex and RoPE score without rCVEs who underwent PFO closure at approximately the same time. The patients with rCVEs and controls participated in a clinical examination, including contrast transthoracic echocardiography (TTE) and Holter electrocardiography, to investigate the possible cause of rCVEs compared with controls. Results: Fourteen (5%) out of the 282 consecutive patients who underwent PFO closure suffered from rCVEs during a mean follow-up of 8.4 years (1.7 rCVEs per 100 patient-years). The median RoPE score of the patients was 7. Recurrent CVE occurred in 3.2 patients per 100 patient-years in patients with residual shunting compared with 0.8 patients per 100 patient-years in those without residual shunt. These patients were on anti-platelet treatment or without any effective anticoagulant treatment at the time rCVE occurred. The risk ratio of rCVEs in patients with residual shunting was 2.9-times higher than in patients without residual shunting (95% CI: 1.4-6.1) at follow-up visit. Four patients who had the BioSTAR device implanted suffered from an rCVE despite lack of residual shunting. Conclusions: This study indicates that residual shunting and choice of the device may be the major reasons for rCVEs.
|
|
| 56. |
- Kvarnbrink, Samuel, 1981-
(författare)
-
LRIG1 in lung cancer : prognostic effects and mechanistic studies
- 2020
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Lung cancer is the leading cause of cancer-related death worldwide as well as in Sweden. Non-small cell lung cancer (NSCLC) is the predominant form, which is largely subdivided into adenocarcinomas and squamous cell carcinomas. A small minority of NSCLC cases that present with localized small tumors are curable with surgery alone, but adjuvant chemotherapy is recommended in most cases that are treated with surgery, even though it only increases the chance of cure by less than 5%. Therefore, additional biomarkers are needed to guide the clinical decision making in early-stage disease.The leucine-rich repeats and immunoglobulin-like (LRIG) family of proteins consists of three paralogous transmembrane proteins that are involved in the regulation of growth factor signaling. Of these proteins, LRIG1 is the most studied and it interacts with a wide variety of growth factor receptors and related proteins, including the epidermal growth factor (EGFR) and several others. High levels of LRIG1 have been associated with better survival in a multitude of malignant diseases, including (but not limited to) breast cancer, bladder cancer, cervical cancer, glioma and melanoma. The aim of this thesis was to investigate whether LRIG1 was a prognostic factor in NSCLC as well, and to further characterize the biological role of LRIG proteins in this disease.To investigate the prognostic impact of LRIG proteins in NSCLC, we stained a tissue microarray (TMA) containing tumor samples from 347 surgically treated early-stage NSCLC patients for LRIG1, LRIG2 and LRIG3. LRIG1 high-expressing adenocarcinoma cases had a large and statistically significant survival benefit of 33 months compared to negative cases. Similarly, an in silico analysis of a large gene expression dataset from the Oncomine database showed that high LRIG1 mRNA expression was linked to better survival as well. Differences in survival persisted even when adjusting for known prognostic factors, meaning that LRIG1 was an independent positive prognostic marker for survival in NSCLC.A yeast two-hybrid screen was performed to search for proteins interacting with a conserved cytosolic peptide shared between all three mammalian LRIG proteins. This screen yielded hits for the paralogous proteins LIM domain only protein 7 (LMO7) and LIM and calponin homology domains-containing protein 1 (LIMCH1). LRIG1 and LRIG3 were both found to physically interact with LMO7 and LIMCH1 as assessed through immunoprecipitation techniques on overexpressed proteins. For LMO7, this could also be confirmed at endogenous protein levels using the proximity ligation assay. The 347 samples in the previously mentioned TMA were stained for LMO7. In our survival analysis, we observed no significant survival differences when looking at LMO7 alone, but the survival benefit observed for high LRIG1 was found to be limited to the subgroup that also was negative for LMO7. This means that the observed physical interaction between LRIG1 and LMO7 appears to translate to differences in patient survival.To investigate possible mechanisms underlying the observed association between high LRIG1 expression and a favorable patient survival, a panel of NSCLC cell lines was modified to overexpress LRIG1. Broadly, LRIG1 overexpression resulted in minor decreases in cellular proliferation rates and no effects on chemosensitivity or radiosensitivity. Looking across the panel of NSCLC cell lines, no clear pattern was observed regarding the effects of LRIG1 overexpression on cellular motility. However, LRIG1 overexpression significantly decreased the clonogenic potential in most cell lines. The only cell line in the panel, H1299, that formed hematogenous disseminated disease in immunodeficient mice was used to establish a mixed-population primary tumor of tagged LRIG1 overexpressing cells and control cells. The LRIG1 transduced cells were was found to be enriched in the injected primary tumor, but no significant changes in their relative abundance was observed between the metastatic sites and their corresponding primary tumors.In summary, we found that LRIG1 was an independent positive prognostic factor in early-stage NSCLC. We identified LMO7 and LIMCH1 as interaction partners for LRIG proteins and showed that the interaction between LMO7 and LRIG1 had implications for the clinical outcome in NSCLC. Furthermore, our mechanistic studies on the effects of LRIG1 overexpression on NSCLC cells suggested that the survival benefit conferred by high LRIG1 expression may be due to differences in metastatic potential. Taken together, the findings in this thesis suggest an important biological role for LRIG proteins in NSCLC.
|
|
| 57. |
- Lévesque, Valérie, et al.
(författare)
-
Implantable Cardioverter-Defibrillators and Patient-Reported Outcomes in Adults with Congenital Heart Disease : an International Study
- 2020
-
Ingår i: Heart Rhythm. - : Elsevier BV. - 1547-5271 .- 1556-3871. ; 17:5, s. 768-776
-
Tidskriftsartikel (refereegranskat)abstract
- BACKGROUND: Implantable cardioverter-defibrillators (ICDs) are increasingly used to prevent sudden deaths in the growing population of adults with congenital heart disease (CHD). Yet, little is known about their impact on patient-reported outcomes (PROs).OBJECTIVE: We assessed and compared PROs in adults with CHD with and without ICDs.METHODS: A propensity-based matching weight analysis was conducted to evaluate PROs in an international cross-sectional study of adults with CHD from 15 countries across 5 continents.RESULTS: A total of 3,188 patients were included: 107 with ICDs and 3,081 weight-matched controls without ICDs. ICD recipients averaged 40.1±12.4 years of age, with >95% having moderate or complex CHD. Defibrillators were implanted for primary and secondary prevention in 38.3% and 61.7%, respectively. Perceived health status, psychological distress, sense of coherence, and health behaviours did not differ significantly in patients with and without ICDs. However, ICD recipients had a more threatening view of their illness (relative % difference 8.56, P=0.011). Those with secondary compared to primary prevention indications had a significantly lower quality of life score (linear analogue scale 72.0±23.1 versus 79.2±13.0, P=0.047). Marked geographic variations were observed. Overall sense of well-being, assessed by a summary score that combines various PROs, was significantly lower in ICD recipients (versus controls) from Switzerland, Argentina, Taiwan, and USA.CONCLUSIONS: In an international cohort of adults with CHD, ICDs were associated with a more threatening illness perception, with a lower quality of life in those with secondary compared to primary prevention indications. However, marked geographic variability in PROs was observed.
|
|
| 58. |
- Lindner, Philip, et al.
(författare)
-
Using alcohol consumption diary data from an internet intervention for outcome and predictive modeling : a validation and machine learning study
- 2020
-
Ingår i: BMC Medical Research Methodology. - : Springer Science and Business Media LLC. - 1471-2288. ; 20:1
-
Tidskriftsartikel (refereegranskat)abstract
- Background: Alcohol use disorder (AUD) is highly prevalent and presents a large treatment gap. Self-help internet interventions are an attractive approach to lowering thresholds for seeking help and disseminating evidence-based programs at scale. Internet interventions for AUD however suffer from high attrition and since continuous outcome measurements are uncommon, little is known about trajectories and processes. The current study investigates whether data from a non-mandatory alcohol consumption diary, common in internet interventions for AUD, approximates drinks reported at follow-up, and whether data from the first half of the intervention predict treatment success.Methods: N = 607 participants enrolled in a trial of online self-help for AUD, made an entry in the non-mandatory consumption diary (total of 9117 entries), and completed the follow-up assessment. Using multiple regression and a subset of calendar data overlapping with the follow-up, scaling factors were derived to account for missing entries per participant and week. Generalized estimating equations with an inverse time predictor were then used to calculate point-estimates of drinks per week at follow-up, the confidence intervals of which were compared to that from the measurement at follow-up. Next, calendar data form the first half of the intervention were retained and summary functions used to create 18 predictors for random forest machine learning models, the classification accuracies of which were ultimately estimated using nested cross-validation.Results: While the raw calendar data substantially underestimated drinks reported at follow-up, the confidence interval of the trajectory-derived point-estimate from the adjusted data overlapped with the confidence interval of drinks reported at follow-up. Machine learning models achieved prediction accuracies of 64% (predicting non-hazardous drinking) and 48% (predicting AUD severity decrease), in both cases with higher sensitivity than specificity.Conclusions: Data from a non-mandatory alcohol consumption diary, adjusted for missing entries, approximates follow-up data at a group level, suggesting that such data can be used to reveal trajectories and processes during treatment and possibly be used to impute missing follow-up data. At an individual level, however, calendar data from the first half of the intervention did not have high predictive accuracy, presumable due to a high rate of missing data and unclear missing mechanisms.
|
|
| 59. |
|
|
| 60. |
- Mai, Vien V., et al.
(författare)
-
Convergence of a stochastic gradient method with momentum for non-smooth non-convex optimization
- 2020
-
Ingår i: 37th International Conference on Machine Learning, ICML 2020. - : International Machine Learning Society (IMLS). ; , s. 6576-6585
-
Konferensbidrag (refereegranskat)abstract
- Stochastic gradient methods with momentum are widely used in applications and at the core of optimization subroutines in many popular machine learning libraries. However, their sample complexities have not been obtained for problems beyond those that are convex or smooth. This paper establishes the convergence rate of a stochastic subgradient method with a momentum term of Polyak type for a broad class of non-smooth, non-convex, and constrained optimization problems. Our key innovation is the construction of a special Lyapunov function for which the proven complexity can be achieved without any tuning of the momentum parameter. For smooth problems, we extend the known complexity bound to the constrained case and demonstrate how the unconstrained case can be analyzed under weaker assumptions than the state-of-The-Art. Numerical results confirm our theoretical developments.
|
|
