Magne Thoresen

• Dalessandro, Matteo & Thoresen, Magne (2026). Methods of Selective Inference for Linear Mixed Models: A Review and Empirical Comparison. International Statistical Review. ISSN 0306-7734. doi: 10.1111/insr.70063. Full text in Research Archive
• Bålsrud, Pia; Ulven, Stine Marie; Hjartåker, Anette; Ottestad, Inger Olaug; Retterstøl, Kjetil & Schwab, Ursula [Show all 9 contributors for this article] (2026). Association between dietary protein intake and frailty index among home-dwelling older adults; an 8-year follow-up study. The Journal of Nutrition, Health & Aging. ISSN 1279-7707. 30(6). doi: 10.1016/j.jnha.2026.100850. Full text in Research Archive
• Skeie, Ivar; Bech, Anne Berit; Clausen, Thomas; Hjemsæter, Arne Jan; Landheim, Anne Signe & Monsbakken, Bent Berntsen [Show all 8 contributors for this article] (2026). Comparison of mortality and causes of death between treatment-engaged individuals with severe alcohol or opioid use disorder: a prospective 19-year clinical cohort study. BMC Psychiatry. 26(1), p. 1–11. doi: 10.1186/s12888-026-07832-5. Full text in Research Archive Show summary

  BACKGROUND: Comparisons of long-term mortality between alcohol-use-disorder (AUD) and opioid-use-disorder (OUD) groups within clinical cohorts are scarce. This study aimed to do this within one clinical cohort comprising groups suffering various severe substance use disorders. METHODS: Design: Observational cohort design. Prospective study of mortality 1998–2016, Cox regression. Sample/setting: 464 persons in specialist SUD treatment at start of observation period. Three subgroups defined by SUD characteristics at the beginning of the observation period, were studied: participants with 1) OUD, on opioid agonist treatment (OAT), (OUD/OAT-group), 2) AUD, no other SUDs (AUD-alone group), 3) AUD as part of a poly-substance-use disorder (AUD/PSUD group). Measures/covariates: Comparison of mortality risks and causes of death (somatic diseases/traumatic causes/substance-induced deaths/unknown cause) between OUD/OAT versus AUD-alone, and AUD-alone versus AUD/PSUD groups. Hazard ratio (HR) adjusted for baseline age (1 January 1998), sex, and socioeconomic/SUD characteristics. RESULTS: AUD-alone versus OUD/OAT: Unadjusted mortality risk was higher in the AUD-alone group [HR 1.99 (95% CI 1.33–2.99)]. However, in adjusted analysis type of SUD showed no statistically significant association with increased mortality [aHR 0.91 (0.54–1.54)], only male sex and increasing baseline age were independently associated with increased mortality risk. Cause-of-death distributions were very similar and somatic diseases caused over 50% of deaths in both groups. AUD-alone versus AUD/PSUD: Unadjusted mortality risk was higher in the AUD-alone group [(HR 2.57 (1.65–4.00)]. However, in adjusted analysis type of SUD showed no statistically significant association with increased mortality [aHR 1.40 (0.81–2.44)], only male sex, increasing baseline age, and “not being in work or education at study inclusion” were independently associated with increased mortality risk. Substance-induced deaths were most common in the AUD/PSUD group (43%). CONCLUSION: In adjusted analyses there were no statistically significant differences in mortality risk between the three subgroups. Somatic diseases were the main cause of death. Besides efforts to improve social rehabilitation and to treat the substance use disorders as such, early and targeted lifestyle interventions, and measures for early detection, treatment and follow-up of somatic disorders are essential to reduce mortality in both AUD and OUD populations. CLINICAL TRIAL NUMBER: Not applicable

• Slettahjell, Hanne Bjørg; Strøm, Vegard; Thoresen, Magne; Strøm, Mona; Glott, Thomas & Biering-Sørensen, Fin [Show all 7 contributors for this article] (2026). Effects of an Individualized Dietitian-Led Nutrition Intervention on Body Composition During the First Year After Traumatic Spinal Cord Injury: A Randomized Controlled Trial. Clinical Nutrition ESPEN. ISSN 2405-4577. doi: 10.1016/j.clnesp.2026.103133. Full text in Research Archive Show summary

  Background & Aims Traumatic spinal cord injury (SCI) induces rapid and adverse changes in body composition, yet effective strategies to prevent excess fat accumulation are limited. This study aimed to evaluate the effect of an individualized dietitian-led nutrition intervention on change in fat mass and related markers of adiposity during the first year following SCI. Methods In this single-center, open-label randomized controlled trial, adults with subacute traumatic SCI were randomized to an individualized dietitian-led intervention or usual care. The primary outcome was change in total fat mass (FM) over one year, assessed by multifrequency bioimpedance analysis. Secondary outcomes included fat mass index (FMI), visceral adipose tissue (VAT), waist circumference, fat-free mass (FFM), fat-free mass index (FFMI), body weight, body mass index (BMI) and biomarkers (fasting glucose, 2-hour oral glucose tolerance test, HbA1c, LDL, HDL and triglycerides). Intention-to-treat analyses used mixed-effects models adjusted for age, sex, and neurological severity. Results Sixty-two participants were randomized (32 intervention, 30 control). Participants were predominantly male (76%) with a mean age of 48 (SD 17) years. At 1 year, the adjusted between-group difference in FM change was −1.5 kg (95% CI −4.8 to 1.9; p = 0.39). Significant between-group differences were observed for VAT (−0.5 L; 95% CI −1.0 to −0.1; p = 0.02) and waist circumference (−4.0 cm; 95% CI −7.0 to −1.0; p = 0.01), with borderline significance for FMI (−1.6 kg/m2; 95% CI −3.2 to 0.0; p = 0.05). No significant between-group differences were observed for body weight, BMI, fat-free mass, or biomarkers. Sensitivity and per-protocol analyses yielded consistent results. Conclusions Although the intervention did not significantly affect the primary endpoint of total fat mass, between-group differences were observed for VAT and waist circumference. These findings suggest that structured, individualized dietetic care initiated during SCI rehabilitation and continued post-discharge may attenuate accumulation of abdominal adiposity. Longer-term follow-up is needed to determine whether early attenuation of abdominal adiposity translates into reduced cardiometabolic risk and improved functional outcomes.

• Bratsberg, Andrea Mathisen; Thoresen, Magne & Goeman, Jelle J. (2025). Bad Estimation, Good Prediction: The Lasso in Dense Regimes. American Statistician. ISSN 0003-1305. doi: 10.1080/00031305.2025.2569464. Full text in Research Archive Show summary

  For high-dimensional omics data, sparsity-inducing regularization methods such as the Lasso are widely used and often yield strong predictive performance, even in settings when the assumption of sparsity is likely violated. We demonstrate that under a specific dense model, namely the high-dimensional joint latent variable model, the Lasso produces sparse prediction rules with favorable prediction error bounds, even when the underlying regression coefficient vector is not sparse at all. We further argue that this model better represents many types of omics data than sparse linear regression models. We prove that the prediction bound under this model in fact decreases with increasing number of predictors, and confirm this through simulation examples. These results highlight the need for caution when interpreting sparse prediction rules, as strong prediction accuracy of a sparse prediction rule may not imply underlying biological significance of the individual predictors.

• Molinari, Marco & Thoresen, Magne (2025). A computationally efficient approach to false discovery rate control and power maximisation via randomisation and mirror statistic. Statistical Methods in Medical Research. ISSN 0962-2802. 34(6), p. 1233–1253. doi: 10.1177/09622802251329768. Full text in Research Archive Show summary

  Simultaneously performing variable selection and inference in high-dimensional regression models is an open challenge in statistics and machine learning. The increasing availability of vast amounts of variables requires the adoption of specific statistical procedures to accurately select the most important predictors in a high-dimensional space, while controlling the false discovery rate (FDR) associated with the variable selection procedure. In this paper, we propose the joint adoption of the Mirror Statistic approach to FDR control, coupled with outcome randomisation to maximise the statistical power of the variable selection procedure, measured through the true positive rate. Through extensive simulations, we show how our proposed strategy allows us to combine the benefits of the two techniques. The Mirror Statistic is a flexible method to control FDR, which only requires mild model assumptions, but requires two sets of independent regression coefficient estimates, usually obtained after splitting the original dataset. Outcome randomisation is an alternative to data splitting that allows to generate two independent outcomes, which can then be used to estimate the coefficients that go into the construction of the Mirror Statistic. The combination of these two approaches provides increased testing power in a number of scenarios, such as highly correlated covariates and high percentages of active variables. Moreover, it is scalable to very high-dimensional problems, since the algorithm has a low memory footprint and only requires a single run on the full dataset, as opposed to iterative alternatives such as multiple data splitting.

• Castel, Charlotte Jocelyne Yvonne; Zhao, Zhi & Thoresen, Magne (2025). Comparing LASSO and IPF-LASSO for multi-modal data: variable selection with Type I error control. Journal of Statistical Computation and Simulation. ISSN 0094-9655. 95(10), p. 2204–2218. doi: 10.1080/00949655.2025.2487503. Full text in Research Archive Show summary

  Variable selection in high-dimensional regression models is challenging. Thus, developing stable and reliable methods for variable selection is essential. Omics data, a common source of high-dimensional data, brings the added complexity of integrating diverse genomic layers into the analysis. The IPF-LASSO model has previously addressed this by applying distinct penalty parameters for each data modality. However, incorporating heterogeneous data layers into variable selection with Type I error control remains an open problem. To address this, we applied stability selection to control the number of false positives in both IPF-LASSO and standard LASSO models. Our study aimed to compare the two methods, investigating whether introducing different penalty parameters per data modality enhances statistical power while controlling false positives. Two high-dimensional data structures were investigated in simulations, one with independent data and the other with correlated data. We also applied the models to breast cancer treatment data, where IPF-LASSO identified relevant clinical variables.

• Bratsberg, Andrea Mathisen; Thoresen, Magne & Ghosh, Abhik (2025). Exponential consistency of M-estimators in generalized linear mixed models. Journal of Statistical Planning and Inference. ISSN 0378-3758. 235. doi: 10.1016/j.jspi.2024.106222. Full text in Research Archive Show summary

  Generalized linear mixed models are powerful tools for analyzing clustered data, where the unknown parameters are classically (and most commonly) estimated by the maximum likelihood and restricted maximum likelihood procedures. However, since the likelihood-based procedures are known to be highly sensitive to outliers, M-estimators have become popular as a means to obtain robust estimates under possible data contamination. In this paper, we prove that for sufficiently smooth general loss functions defining the M-estimators in generalized linear mixed models, the tail probability of the deviation between the estimated and the true regression coefficients has an exponential bound. This implies an exponential rate of consistency of these M-estimators under appropriate assumptions, generalizing the existing exponential consistency results from univariate to multivariate responses. We have illustrated this theoretical result further for the special examples of the maximum likelihood estimator and the robust minimum density power divergence estimator, a popular example of model-based M-estimators, in the settings of linear and logistic mixed models, comparing it with the empirical rate of convergence through simulation studies.

• Krutto, Annika; Nøst, Therese Haugdahl & Thoresen, Magne (2024). A heavy-tailed model for analyzing miRNA-seq raw read counts. Statistical Applications in Genetics and Molecular Biology. 23(1). doi: 10.1515/sagmb-2023-0016. Full text in Research Archive Show summary

  This article addresses the limitations of existing statistical models in analyzing and interpreting highly skewed miRNA-seq raw read count data that can range from zero to millions. A heavy-tailed model using discrete stable distributions is proposed as a novel approach to better capture the heterogeneity and extreme values commonly observed in miRNA-seq data. Additionally, the parameters of the discrete stable distribution are proposed as an alternative target for differential expression analysis. An R package for computing and estimating the discrete stable distribution is provided. The proposed model is applied to miRNA-seq raw counts from the Norwegian Women and Cancer Study (NOWAC) and the Cancer Genome Atlas (TCGA) databases. The goodness-of-fit is compared with the popular Poisson and negative binomial distributions, and the discrete stable distributions are found to give a better fit for both datasets. In conclusion, the use of discrete stable distributions is shown to potentially lead to more accurate modeling of the underlying biological processes.

• Olsen, Thomas; Stolt, Emma Theresia Kjelland; Øvrebø, Bente; Elshorbagy, Amany; Tore, Elena C. & Lee-Ødegård, Sindre [Show all 20 contributors for this article] (2024). Dietary sulfur amino acid restriction in humans with overweight and obesity: a translational randomized controlled trial. Journal of Translational Medicine. 22(40). doi: 10.1186/s12967-023-04833-w. Full text in Research Archive Show summary

  Background Dietary sulfur amino acid restriction (SAAR) improves metabolic health in animals. In this study, we investigated the effect of dietary SAAR on body weight, body composition, resting metabolic rate, gene expression profiles in white adipose tissue (WAT), and an extensive blood biomarker profile in humans with overweight or obesity. Methods N = 59 participants with overweight or obesity (73% women) were randomized stratified by sex to an 8-week plant-based dietary intervention low (~ 2 g/day, SAAR) or high (~ 5.6 g/day, control group) in sulfur amino acids. The diets were provided in full to the participants, and both investigators and participants were blinded to the intervention. Outcome analyses were performed using linear mixed model regression adjusted for baseline values of the outcome and sex. Results SAAR led to a ~ 20% greater weight loss compared to controls (β 95% CI − 1.14 (− 2.04, − 0.25) kg, p = 0.013). Despite greater weight loss, resting metabolic rate remained similar between groups. Furthermore, SAAR decreased serum leptin, and increased ketone bodies compared to controls. In WAT, 20 genes were upregulated whereas 24 genes were downregulated (FDR < 5%) in the SAAR group compared to controls. Generally applicable gene set enrichment analyses revealed that processes associated with ribosomes were upregulated, whereas processes related to structural components were downregulated. Conclusion Our study shows that SAAR leads to greater weight loss, decreased leptin and increased ketone bodies compared to controls. Further research on SAAR is needed to investigate the therapeutic potential for metabolic conditions in humans. Trial registration: ClinicalTrials.gov identifier: NCT04701346, registered Jan 8th 2021, https://www.clinicaltrials.gov/study/NCT04701346

• Olsen, Thomas; Vinknes, Kathrine Jørgensen; Barvíková, Kristýna; Stolt, Emma Theresia Kjelland; Lee-Ødegård, Sindre & Troensegaard, Hannibal [Show all 18 contributors for this article] (2024). Dietary sulfur amino acid restriction in humans with overweight and obesity: Evidence of an altered plasma and urine sulfurome, and a novel metabolic signature that correlates with loss of fat mass and adipose tissue gene expression. Redox Biology. 73. doi: 10.1016/j.redox.2024.103192. Full text in Research Archive Show summary

  Background In animals, dietary sulfur amino acid restriction (SAAR) improves metabolic health, possibly mediated by altering sulfur amino acid metabolism and enhanced anti-obesogenic processes in adipose tissue. Aim To assess the effects of SAAR over time on the plasma and urine SAA-related metabolites (sulfurome) in humans with overweight and obesity, and explore whether such changes were associated with body weight, body fat and adipose tissue gene expression. Methods Fifty-nine subjects were randomly allocated to SAAR (∼2 g SAA, n = 31) or a control diet (∼5.6 g SAA, n = 28) consisting of plant-based whole-foods and supplemented with capsules to titrate contents of SAA. Sulfurome metabolites in plasma and urine at baseline, 4 and 8 weeks were measured using HPLC and LC-MS/MS. mRNA-sequencing of subcutaneous white adipose tissue (scWAT) was performed to assess changes in gene expression. Data were analyzed with mixed model regression. Principal component analyses (PCA) were performed on the sulfurome data to identify potential signatures characterizing the response to SAAR. Results SAAR led to marked decrease of the main urinary excretion product sulfate (p < 0.001) and plasma and/or 24-h urine concentrations of cystathionine, sulfite, thiosulfate, H2S, hypotaurine and taurine. PCA revealed a distinct metabolic signature related to decreased transsulfuration and H2S catabolism that predicted greater weight loss and android fat mass loss in SAAR vs. controls (all pinteraction < 0.05). This signature correlated positively with scWAT expression of genes in the tricarboxylic acid cycle, electron transport and β-oxidation (FDR = 0.02). Conclusion SAAR leads to distinct alterations of the plasma and urine sulfurome in humans, and predicted increased loss of weight and android fat mass, and adipose tissue lipolytic gene expression in scWAT. Our data suggest that SAA are linked to obesogenic processes and that SAAR may be useful for obesity and related disorders.

• Das, Siddhartha; Rundblad, Amanda; Marques, Irene Fontes; Soares, Ana Goncalves; Jaddoe, Vincent W. & Vrijheid, Martine [Show all 14 contributors for this article] (2024). Air pollution exposure is associated with gene expression in children. Environmental Epigenetics. 10(1). doi: 10.1093/eep/dvae025. Full text in Research Archive Show summary

  Environmental exposures, including air pollutants and lack of natural spaces, are associated with suboptimal health outcomes in children. We aimed to study the associations between environmental exposures and gene expression in children. Associations of exposure to particulate matter (PM) with diameter <2.5 (PM2.5) and < 10 (PM10) micrometers, nitrogen dioxide, green spaces, and blue space, with whole-blood gene expression were explored in children from the Dutch Generation R Study (n = 172). Analyses were adjusted for age, sex, batch, maternal education, and area socioeconomic status. Follow-up analysis was carried out using lymphoblastoid cell line gene expression in children from the ALSPAC Study (n = 946). Gene set enrichment analysis (GSEA) using hallmark and immune gene sets from the molecular signature database was carried out to identify significantly over-represented gene sets for insights into biological mechanisms Exposure to PM2.5 was associated with expression of 86 genes in discovery analyses in the Generation R Study [false discovery rate (FDR)-adjusted P-value < .25]. Of these, PM2.5 was also associated with GNG11 expression in the same direction in follow-up analysis (FDR-adjusted P-value < .05). The remaining exposures showed much fewer associations in the discovery analyses. GSEA using PM2.5 association results for both cohorts indicated suppression of gene sets related to interferon response and response to bacterial and viral exposure. In conclusion, gene expression analysis performed in two independent cohorts suggests that PM2.5 exposure in children may be involved in interferon and microbial infection responses.

• Hansen, Christin Saltnes; Rustøen, Tone; Beitland, Sigrid; Puntillo, Kathleen; Thoresen, Magne & Hofsø, Kristin (2024). Subgroups of ICU patients identified by self-reported symptoms - A prospective multicenter study. Intensive and Critical Care Nursing. ISSN 0964-3397. 84, p. 1–8. doi: 10.1016/j.iccn.2024.103761. Full text in Research Archive Show summary

  Background: Intensive care unit (ICU) patients experience several symptoms, yet patterns of symptoms and their relationship with demographic and clinical characteristics have not previously been investigated. Objectives: To identify and compare subgroups (i.e. latent symptom classes) of intensive ICU patients based on prevalence of co-occurring symptoms over seven days. Research methodology: Prospective cohort study of adult ICU patients’ self-reports of five symptoms during seven days in ICU. Latent class analysis was applied to identify subgroups of ICU patients. Setting: Multicenter study with patients from six mixed ICUs in Norway. Main Outcome Measures: Patient Symptom Survey was used to assess five symptoms (i.e., thirst, pain, anxiousness, tiredness, shortness of breath). Results: Among 353 included patients, median age was 63 years and 60.3 % were male. Subgroups of patients were identified in a Low class (n = 126, 35.7 %), Middle Class (n = 177, 50.1 %) and High Class (n = 50, 14.2 %) based on reporting of the prevalence of five symptoms. Patients in the Low class had a low prevalence of all symptoms. Middle Class patients had a high prevalence of thirst and tiredness and a low prevalence of pain, anxiousness and shortness of breath. The High class patients had a high prevalence of all symptoms. Symptom prevalence remained stable in the Low and Middle class over time and increased over time in the High class. There were significant differences among symptom classes in use of mechanical ventilation (p = 0.012), analgesics (p < 0.001), alpha-2 agonists (p = 0.004) and fluid restriction (p = 0.006). Patients in the High class received more of these ICU-treatments. Conclusions: Findings suggest that subgroups of ICU patients with distinct symptom experiences can be identified. The High prevalence class patients had consistently high levels of all symptoms across seven ICU days and received more ICU-related interventions. Implication for clinical practice: Some ICU patients experience a consistently high prevalence of co-occurring symptoms. Clinicians should be aware of treatment factors that could be linked to a high burden of symptoms.

• Djordjilovic, Vera; Ponzi, Erica; Nøst, Therese Haugdahl & Thoresen, Magne (2024). penalizedclr: an R package for penalized conditional logistic regression for integration of multiple omics layers. BMC Bioinformatics. 25(1). doi: 10.1186/s12859-024-05850-2. Full text in Research Archive Show summary

  Background: The matched case–control design, up until recently mostly pertinent to epidemiological studies, is becoming customary in biomedical applications as well. For instance, in omics studies, it is quite common to compare cancer and healthy tissue from the same patient. Furthermore, researchers today routinely collect data from various and variable sources that they wish to relate to the case–control status. This highlights the need to develop and implement statistical methods that can take these tendencies into account. Results: We present an R package penalizedclr, that provides an implementation of the penalized conditional logistic regression model for analyzing matched case–control studies. It allows for different penalties for different blocks of covariates, and it is therefore particularly useful in the presence of multi-source omics data. Both L1 and L2 penalties are implemented. Additionally, the package implements stability selection for variable selection in the considered regression model. Conclusions: The proposed method fills a gap in the available software for fitting high-dimensional conditional logistic regression models accounting for the matched design and block structure of predictors/features. The output consists of a set of selected variables that are significantly associated with case–control status. These variables can then be investigated in terms of functional interpretation or validation in further, more targeted studies.

• Johansen, Anja Katrin; Bogsrud, Martin Prøven; Thoresen, Magne; Christensen, Jacob J.; Narverud, Ingunn & Langslet, Gisle [Show all 9 contributors for this article] (2024). Lipoprotein(a) in children and adolescents with genetically confirmed familial hypercholesterolemia followed up at a specialized lipid clinic. Atherosclerosis Plus. ISSN 2667-0909. 57, p. 13–18. doi: 10.1016/j.athplu.2024.06.002. Full text in Research Archive Show summary

  Background and aim Many children with an FH mutation also exhibit elevated lipoprotein(a) levels, which is an independent risk factor for atherosclerotic cardiovascular disease. Studies have reported higher levels of lipoprotein(a) in adult and middle-aged women than men. There is limited knowledge on the concentration and change of lipoprotein(a) levels in children with genetic FH, and therefore we investigated sex-differences in lipoprotein(a) level and change in lipoprotein(a) in girls and boys with genetically confirmed FH. Methods Medical records were reviewed retrospectively in 438 subjects with heterozygous FH that started follow-up below the age of 19 years at the Lipid Clinic, Oslo University Hospital in Norway, and of these we included 386 subjects with at least one Lp(a) measurement. Results Mean (SD) age at baseline was 13.8 (7.3) years and the age was similar between sexes. Girls had a higher lipoprotein(a) level than boys at baseline: median (25–75 percentile) 223 (108–487) vs. 154 (78–360) mg/L, respectively (p < 0.01). From baseline to follow-up measurement (mean [SD] 8.9 [6.1] years apart), the mean (95 % CI) absolute and percentage change in Lp(a) level in girls was 151.4 (54.9–247.8) mg/L and 44.8 (16.4–73.1) %, respectively, and in boys it was 66.8 (22.9–110.8) mg/L and 50.5 (8.8–92.3) %, respectively (both p > 0.05). Conclusions We found an increase in Lp(a) levels in children with genetic FH with age, and higher levels in girls than boys, which could impact risk assessment and future ASCVD. Further research is needed to elucidate whether subjects with FH could benefit from lipoprotein(a)-lowering therapies that are under current investigations.

• Dalum, Helene Seljenes; Tyssen, Reidar; Moum, Torbjørn; Thoresen, Magne & Hem, Erlend (2024). Euthanasia of animals – association with veterinarians’ suicidal thoughts and attitudes towards assisted dying in humans: a nationwide cross-sectional survey (the NORVET study). BMC Psychiatry. 24(2). doi: 10.1186/s12888-023-05402-7. Full text in Research Archive Show summary

  Background Veterinarians are an occupational group with an increased suicide risk. Euthanasing animals may influence both veterinarians’ views on assisted dying in humans and their suicide risk. We investigated (I) attitudes towards assisted dying, (II) whether the field of work and the frequency of euthanasing animals were associated with positive attitudes towards human euthanasia, and (III) whether frequently euthanasing animals was associated with serious suicidal thoughts. Methods We conducted a nationwide cross-sectional study among veterinarians in Norway (response rate: 75%). Logistic regression models were used to calculate the odds ratios for both positive attitudes towards human euthanasia and serious suicidal thoughts. The analyses were adjusted for socio-demographic and work-related factors. Results Fifty-five percent of the veterinarians agreed that euthanasia should be permitted for humans with a fatal disease and short life expectancy. Working with companion animals was independently associated with positive attitudes towards human euthanasia (OR = 1.66 (95% CI: 1.23–2.23)), while veterinarians’ frequency of euthanasing animals was not. Frequency of euthanasing animals was independently associated with serious suicidal thoughts, OR = 2.56 (95% CI: 1.35–4.87). Conclusions Veterinarians’ attitudes towards assisted dying in humans did not differ from those of the general population. Veterinarians’ frequency of euthanasing animals was not associated with positive attitudes towards euthanasia in humans. However, veterinarians working in companion animal practices were more likely to have positive attitudes towards euthanasia in humans. Moreover, euthanising animals five times or more a week was associated with serious suicidal thoughts. We need more research to infer about causality in these findings.

• Herfindal, Anne Mari; Nilsen, Morten; Aspholm, Trude Elise; Schultz, Gry Irene Granli; Valeur, Jørgen & Rudi, Knut [Show all 10 contributors for this article] (2024). Effects of fructan and gluten on gut microbiota in individuals with self-reported non-celiac gluten/wheat sensitivity—a randomised controlled crossover trial. BMC Medicine. 22:358(1), p. 1–19. doi: 10.1186/s12916-024-03562-1. Full text in Research Archive Show summary

  Background Individuals with non-celiac gluten/wheat sensitivity (NCGWS) experience improvement in gastrointestinal symptoms following a gluten-free diet. Although previous results have indicated that fructo-oligosaccharides (FOS), a type of short-chain fructans, were more likely to induce symptoms than gluten in self-reported NCGWS patients, the underlying mechanisms are unresolved. Methods Our main objective was therefore to investigate whether FOS-fructans and gluten affect the composition and diversity of the faecal microbiota (16S rRNA gene sequencing), faecal metabolites of microbial fermentation (short-chain fatty acids [SCFA]; gas chromatography with flame ionization detector), and a faecal biomarker of gut inflammation (neutrophil gelatinase-associated lipocalin, also known as lipocalin 2, NGAL/LCN2; ELISA). In the randomised double-blind placebo-controlled crossover study, 59 participants with self-reported NCGWS underwent three different 7-day diet challenges with gluten (5.7 g/day), FOS-fructans (2.1 g/day), and placebo separately (three periods, six challenge sequences). Results The relative abundances of certain bacterial taxa were affected differently by the diet challenges. After the FOS-fructan challenge, Fusicatenibacter increased, while Eubacterium (E.) coprostanoligenes group, Anaerotruncus, and unknown Ruminococcaceae genera decreased. The gluten challenge was primarily characterized by increased abundance of Eubacterium xylanophilum group. However, no differences were found for bacterial diversity (α-diversity), overall bacterial community structure (β-diversity), faecal metabolites (SCFA), or NGAL/LCN2. Furthermore, gastrointestinal symptoms in response to FOS-fructans were generally not linked to substantial shifts in the gut bacterial community. However, the reduction in E. coprostanoligenes group following the FOS-fructan challenge was associated with increased gastrointestinal pain. Finally, correlation analysis revealed that changes in gastrointestinal symptoms following the FOS-fructan and gluten challenges were linked to varying bacterial abundances at baseline. Conclusions In conclusion, while FOS-fructans induced more gastrointestinal symptoms than gluten in the NCGWS patients, we did not find that substantial shifts in the composition nor function of the faecal microbiota could explain these differences in the current study. However, our results indicate that individual variations in baseline bacterial composition/function may influence the gastrointestinal symptom response to both FOS-fructans and gluten. Additionally, the change in E. coprostanoligenes group, which was associated with increased symptoms, implies that attention should be given to these bacteria in future trials investigating the impact of dietary treatments on gastrointestinal symptoms. Trial registration Clinicaltrials.gov as NCT02464150.

• Uchai, Shreeshti; Andersen, Lene Frost; Thoresen, Magne; Hopstock, Laila Arnesdatter & Hjartåker, Anette (2024). Does the association between adiposity measures and pre-frailty among older adults vary by social position? Findings from the Tromsø study 2015/2016. BMC Public Health. 24(1). doi: 10.1186/s12889-024-18939-3. Full text in Research Archive Show summary

  Introduction - Pre-frailty provides an ideal opportunity to prevent physical frailty and promote healthy ageing. Excess adiposity has been associated with an increased risk of pre-frailty, but limited studies have explored whether the association between adiposity measures and pre-frailty varies by social position.<p> <p>Methods - We used data from the seventh survey of the Tromsø Study (Tromsø7) conducted in 2015–2016. Our primary sample consisted of 2,945 women and 2,794 men aged ≥ 65 years. Pre-frailty was defined as the presence of one or two of the five frailty components: low grip strength, slow walking speed, exhaustion, unintentional weight loss and low physical activity. Adiposity was defined by body mass index (BMI), waist circumference (WC), fat mass index (FMI) and visceral adipose tissue (VAT) mass. Education and subjective social position were used as measures of social position. Poisson regression with robust variance was used to assess the association between adiposity measures and pre-frailty, and the interaction term between adiposity measures and social position measures were utilised to explore whether the association varied by social position.<p> <p>Results - In our sample, 28.7% of women and 25.5% of men were pre-frail. We found sub-multiplicative interaction of BMI-defined obesity with education in women and subjective social position in men with respect to development of pre-frailty. No other adiposity measures showed significant variation by education or subjective social position. Regardless of the levels of education or subjective social position, participants with excess adiposity (high BMI, high WC, high FMI and high VAT mass) had a higher risk of pre-frailty compared to those with low adiposity.<p> <p>Conclusion - We consistently observed that women and men with excess adiposity had a greater risk of pre-frailty than those with low adiposity, with only slight variation by social position. These results emphasize the importance of preventing excess adiposity to promote healthy ageing and prevent frailty among all older adults across social strata.

• Bratsberg, Andrea Mathisen; Ghosh, Abhik & Thoresen, Magne (2024). Conditional Variable Screening for Ultra-High Dimensional Longitudinal Data With Time Interactions. Biometrical Journal. ISSN 0323-3847. 66(8), p. 1–15. doi: 10.1002/bimj.70005. Full text in Research Archive Show summary

  In recent years, we have been able to gather large amounts of genomic data at a fast rate, creating situations where the number of variables greatly exceeds the number of observations. In these situations, most models that can handle a moderately high dimension will now become computationally infeasible or unstable. Hence, there is a need for a prescreening of variables to reduce the dimension efficiently and accurately to a more moderate scale. There has been much work to develop such screening procedures for independent outcomes. However, much less work has been done for high-dimensional longitudinal data in which the observations can no longer be assumed to be independent. In addition, it is of interest to capture possible interactions between the genomic variable and time in many of these longitudinal studies. In this work, we propose a novel conditional screening procedure that ranks variables according to the likelihood value at the maximum likelihood estimates in a marginal linear mixed model, where the genomic variable and its interaction with time are included in the model. This is to our knowledge the first conditional screening approach for clustered data. We prove that this approach enjoys the sure screening property, and assess the finite sample performance of the method through simulations.

• Konglevoll, Dina Moxness; Andersen, Lene Frost; Thoresen, Magne; Totland, Torunn Holm; Hopstock, Laila Arnesdatter & Hjartåker, Anette [Show all 7 contributors for this article] (2024). Dietary trajectories over 21 years and frailty in Norwegian older adults: the Tromsø Study 1994–2016. European Journal of Nutrition. ISSN 1436-6207. 63, p. 2987–2998. doi: 10.1007/s00394-024-03482-z. Full text in Research Archive Show summary

  Purpose To investigate the association between five dietary trajectories over 21 years and frailty in Norwegian older adults.<p> <p>Methods This study used data from three surveys of the Tromsø Study. Diet was measured using food frequency questionnaires at baseline (Tromsø4, 1994–95), after 7 years (Tromsø5, 2001) and at the end of follow-up (Tromsø7, 2015–16). Survey-specific diet scores were constructed based on the Nordic Nutrition Recommendations 2023 and group-based trajectory modelling was used to derive dietary trajectories. At follow-up, frailty was assessed with a 41-item frailty index. Linear regression analysis was performed to assess the associations between dietary trajectories and frailty, adjusted for baseline variables. <p>Results Among the 715 participants, 55% were women, with an average age of 54 years at baseline and 74 years at followup. The dietary trajectories ‘moderately healthy’ and ‘healthy increase’ were associated with a lower frailty index score at follow-up (β = −0.02, 95% confidence interval (CI)=−0.04, −0.002, β = −0.03, 95% CI=−0.06, −0.007), compared with the ‘unhealthy’ trajectory. <p>Conclusion Our findings suggest that maintaining a moderately healthy to very healthy diet from mid-life into older age is associated with a lower risk of frailty and supports the promotion of a healthy diet from adulthood to facilitate healthy ageing.

• Øyri, Linn K. L.; Christensen, Jacob J.; Sebert, Sylvain; Thoresen, Magne; Michelsen, Trond M. & Ulven, Stine M. [Show all 12 contributors for this article] (2023). Maternal prenatal cholesterol levels predict offspring weight trajectories during childhood in the Norwegian Mother, Father and Child Cohort Study. BMC Medicine. 21. doi: 10.1186/s12916-023-02742-9. Full text in Research Archive Show summary

  Background Numerous intrauterine factors may affect the offspring’s growth during childhood. We aimed to explore if maternal and paternal prenatal lipid, apolipoprotein (apo)B and apoA1 levels are associated with offspring weight, length, and body mass index from 6 weeks to eight years of age. This has previously been studied to a limited extent. Methods This parental negative control study is based on the Norwegian Mother, Father and Child Cohort Study and uses data from the Medical Birth Registry of Norway. We included 713 mothers and fathers with or without self-reported hypercholesterolemia and their offspring. Seven parental metabolites were measured by nuclear magnetic resonance spectroscopy, and offspring weight and length were measured at 12 time points. Data were analyzed by linear spline mixed models, and the results are presented as the interaction between parental metabolite levels and offspring spline (age). Results Higher maternal total cholesterol (TC) level was associated with a larger increase in offspring body weight up to 8 years of age (0.03 ≤ Pinteraction ≤ 0.04). Paternal TC level was not associated with change in offspring body weight (0.17 ≤ Pinteraction ≤ 0.25). Higher maternal high-density lipoprotein cholesterol (HDL-C) and apoA1 levels were associated with a lower increase in offspring body weight up to 8 years of age (0.001 ≤ Pinteraction ≤ 0.005). Higher paternal HDL-C and apoA1 levels were associated with a lower increase in offspring body weight up to 5 years of age but a larger increase in offspring body weight from 5 to 8 years of age (0.01 ≤ Pinteraction ≤ 0.03). Parental metabolites were not associated with change in offspring height or body mass index up to 8 years of age (0.07 ≤ Pinteraction ≤ 0.99). Conclusions Maternal compared to paternal TC, HDL-C, and apoA1 levels were more strongly and consistently associated with offspring body weight during childhood, supporting a direct intrauterine effect.

• Romeo, Giovanni; Buonaccorsi, John P. & Thoresen, Magne (2023). Detecting and correcting for heteroscedasticity in the presence of measurement error. Communications in Statistics - Simulation and Computation. ISSN 0361-0918. doi: 10.1080/03610918.2023.2190061. Full text in Research Archive Show summary

  The problem of measurement error in predictors has received a vast amount of attention for both linear and non-linear models. The major focus has generally been on inference for regression coefficients, while residual analysis, variance estimation and heteroscedasticity assessment have received limited attention. In this paper we propose and evaluate new methods, based on reconstructed predictors, to address these problems in the context of linear regression. Methods using moment reconstructed predictors are found to perform well, including a new model-based bootstrapping technique for testing heteroscedasticity. Motivated by an example where the predictor is methylation rate and the measurement error model is based on the binomial distribution, particular attention is given to being able to handle the situation where both the measurement error variances and the variance of the error in the equation may be changing as a quadratic function of the unobserved predictor.

• Konglevoll, Dina Moxness; Andersen, Lene Frost; Hopstock, Laila Arnesdatter; Strand, Bjørn Heine; Thoresen, Magne & Totland, Torunn Holm [Show all 8 contributors for this article] (2023). Fish intake and pre-frailty in Norwegian older adults - a prospective cohort study: the Tromsø Study 1994–2016. BMC Geriatrics. 23(1). doi: 10.1186/s12877-023-04081-z. Full text in Research Archive Show summary

  Background Pre-frailty is an intermediate, potentially reversible state before the onset of frailty. Healthy dietary choices may prevent pre-frailty. Fish is included in most healthy diets, but little is known about the association between long-term habitual fish intake and pre-frailty. We aimed to elucidate the longitudinal association between the frequency of fish intake and pre-frailty in a cohort of older adults in Norway.<p> <p>Methods 4350 participants (52% women,≥65 years at follow-up) were included in this prospective cohort study. Data was obtained from three waves of the population-based Tromsø Study in Norway; Tromsø4 (1994–1995), Tromsø6 (2007–2008) and Tromsø7 (follow-up, 2015–2016). Frailty status at follow-up was defined by a modifed version of Fried’s phenotype. Fish intake was self-reported in the three surveys and assessed as three levels of frequency of intake: low (0–3 times/month), medium (1–3 times/week) and high (≥4 times/week). The fsh–pre-frailty association was analysed using multivariable logistic regression in two ways; (1) frequency of intake of lean, fatty and total fish in Tromsø6 and pre-frailty at follow-up, and (2) patterns of total fsh intake across the three surveys and pre-frailty at follow-up. <p>Results At follow-up, 28% (n=1124) were pre-frail. Participants with a higher frequency of lean, fatty and total fish intake had 28% (odds ratio (OR)=0.72, 95% confidence interval (CI)=0.53, 0.97), 37% (OR=0.63, 95% CI=0.43, 0.91) and 31% (OR=0.69, 95% CI=0.52, 0.91) lower odds of pre-frailty 8 years later compared with those with a low intake, respectively. A pattern of stable high fish intake over 21 years was associated with 41% (OR=0.59, 95% CI=0.38, 0.91) lower odds of pre-frailty compared with a stable low intake. <p>Conclusions A higher frequency of intake of lean, fatty and total fish, and a pattern of consistent frequent fish intake over time, were associated with lower odds of pre-frailty in older community-dwelling Norwegian adults. These results emphasise the important role of fish in a healthy diet and that a frequent fish intake should be promoted to facilitate healthy ageing.

• Hofmann, Bjørn Morten; Thoresen, Magne & Holm, Søren (2023). Research Integrity Attitudes and Behaviors are Difficult to alter: Results from a ten Year Follow-up Study in Norway. Journal of Empirical Research on Human Research Ethics. ISSN 1556-2646. 18(1/2), p. 50–57. doi: 10.1177/15562646221150032. Full text in Research Archive Show summary

  Background: Research integrity has obtained much attention in research communities, but also in the general public. To improve research integrity is difficult as it involves complex systems of knowledge, attitudes, and practices. The objective of this study is to investigate the knowledge, attitudes, and practices of cohorts of PhD candidates at one faculty (of medicine) over time and compare this to finished PhDs of the same cohorts. Material and method: Researchers (n  =  186) awarded the degree PhD at the Faculty of Medicine at the University of Oslo in 2019 were invited to answer a questionnaire about knowledge, attitudes and actions related to scientific dishonesty. 94 responded (50.5%). The results were compared with results among first-year PhD candidates who responded to the same questionnaire during 2010–20 (n  =  536) and to those who finished PhDs in 2016 (n  =  86). Results: For the years 2010–2020 1.1% of the PhD candidates report to have engaged in severe scientific misconduct (FFP) while 0.9% report to have presented results in a misleading way. 2.3% report that they know of persons at their department who have engaged in FFP the last 12 months. In total 1.5% report to have experienced pressure to engage in severe scientific misconduct (FFP) while 2.1% report to have experienced pressure to present results in a misleading way. On average 12.8% report to have been exposed to unethical pressure concerning inclusion or ordering of authors during the last 12 months, and 28.8% report to have knowledge about their department's written policies about research integrity. While some attitudes improve over the years, attitudes in general are not much changed from 2010–2020. None of the PhDs that received a PhD from the Faculty of Medicine at the University of Oslo in 2019 reported to have engaged in FFT or having experienced pressure to do so.1.1% experienced pressure to present results in other misleading ways, while 26.6% of respondents had experienced unethical pressure in relation to authorship during the course of the PhD fellowship. 4.3% knew about someone at their department who had presented results in a misleading manner. Some attitudes were not in line with traditional conceptions of research integrity, but most agreed that their research environment displayed research integrity. Conclusion: This long-term follow up study shows that few PhD-candidates report to engage in severe scientific misconduct, that they experience little pressure to do so, and with some exceptions, attitudes in in line with good research integrity. However, pressure in relation to authorship is relatively common. There is some improvement in research integrity from PhD candidates to recently finished PhDs, but in general research integrity is stable over time.

• Ghosh, Abhik; Ponzi, Erica; Sandanger, Torkjel M & Thoresen, Magne (2022). Robust sure independence screening for nonpolynomial dimensional generalized linear models. Scandinavian Journal of Statistics. ISSN 0303-6898. doi: 10.1111/sjos.12628. Full text in Research Archive Show summary

  We consider the problem of variable screening in ultra-high-dimensional generalized linear models (GLMs) of nonpolynomial orders. Since the popular SIS approach is extremely unstable in the presence of contamination and noise, we discuss a new robust screening procedure based on the minimum density power divergence estimator (MDPDE) of the marginal regression coefficients. Our proposed screening procedure performs well under pure and contaminated data scenarios. We provide a theoretical motivation for the use of marginal MDPDEs for variable screening from both population as well as sample aspects; in particular, we prove that the marginal MDPDEs are uniformly consistent leading to the sure screening property of our proposed algorithm. Finally, we propose an appropriate MDPDE-based extension for robust conditional screening in GLMs along with the derivation of its sure screening property. Our proposed methods are illustrated through extensive numerical studies along with an interesting real data application.

• Dalum, Helene Seljenes; Tyssen, Reidar; Moum, Torbjørn Åge; Thoresen, Magne & Hem, Erlend (2022). Professional help-seeking behaviour for mental health problems among veterinarians in Norway: a nationwide, cross-sectional study (The NORVET study). BMC Public Health. 22. doi: 10.1186/s12889-022-13710-y. Full text in Research Archive Show summary

  Background Veterinarians have a relatively high prevalence of mental health problems; however, research on professional help-seeking is limited. The main purpose of the present study was to investigate the prevalence of mental health problems and professional help-seeking behaviour for such problems, and the independent factors associated with help-seeking behaviour among veterinarians in Norway. Method This cross-sectional study included all veterinarians in Norway (response rate 75%, 70% women). Logistic regression was used to calculate odds ratios (OR) for professional help-seeking for mental health problems. Analyses were controlled for socio-demographic, individual (personality trait reality weakness, SCL-5, attitudes toward mental illness), and work-related factors (work field, job stress). Results The prevalence of self-reported mental health problems in need of treatment was 30% (746/2494), significantly higher among women than men (36% vs. 15%). Fifty-four percent had sought professional help, women significantly more often (56%) than men (41%). Among veterinarians with serious suicidal thoughts, 50% (69/139) had sought help. Veterinarians most frequently related mental health problems to work problems (47%), women significantly more often (49%) than men (34%). Factors significantly associated with help-seeking were being female, OR = 2.11 (95% CI: 1.24–3.60), working with production animals, OR = 0.35 (0.13–0.98), public administration, OR = 2.27 (1.15–4.45), academia/research, OR = 4.78 (1.99–11.47) or ‘other’ fields, OR = 2.79 (1.23–6.32), and attitudes toward mental illness, OR = 1.32 (1.03–1.68). Conclusions Thirty percent of veterinarians in Norway reported mental health problems in need of treatment, and only half of them had sought professional help. A low degree of help-seeking was also seen among those with serious suicidal thoughts. Being female, positive attitudes toward treatment of mental illness, working in public administration, academia/research and ‘other’ field were associated with more help-seeking, while working in production animal practice was associated with less help-seeking. Interventions to increase help-seeking behaviour for mental health problems among veterinarians are warranted.

• Page, Christian Magnus; Nøst, Therese Haugdahl; Djordjilović, Vera; Thoresen, Magne; Frigessi, Arnoldo & Sandanger, Torkjel M [Show all 7 contributors for this article] (2022). Pre-diagnostic DNA methylation in blood leucocytes in cutaneous melanoma; a nested case–control study within the Norwegian Women and Cancer cohort. Scientific Reports. 12(1). doi: 10.1038/s41598-022-18585-y. Full text in Research Archive Show summary

  The prognosis of cutaneous melanoma depends on early detection, and good biomarkers for melanoma risk may provide a valuable tool to detect melanoma development at a pre-clinical stage. By studying the epigenetic profile in pre-diagnostic blood samples of melanoma cases and cancer free controls, we aimed to identify DNA methylation sites conferring melanoma risk. DNA methylation was measured at 775,528 CpG sites using the Illumina EPIC array in whole blood in incident melanoma cases (n = 183) and matched cancer-free controls (n = 183) in the Norwegian Women and Cancer cohort. Phenotypic information and ultraviolet radiation exposure were obtained from questionnaires. Epigenome wide association (EWAS) was analyzed in future melanoma cases and controls with conditional logistic regression, with correction for multiple testing using the false discovery rate (FDR). We extended the analysis by including a public data set on melanoma (GSE120878), and combining these different data sets using a version of covariate modulated FDR (AdaPT). The analysis on future melanoma cases and controls did not identify any genome wide significant CpG sites (0.85 ≤ padj ≤ 0.99). In the restricted AdaPT analysis, 7 CpG sites were suggestive at the FDR level of 0.15. These CpG sites may potentially be used as pre-diagnostic biomarkers of melanoma risk.

• Djordjilović, Vera; Hemerik, Jesse & Thoresen, Magne (2022). On optimal two-stage testing of multiple mediators. Biometrical Journal. ISSN 0323-3847. 64(6), p. 1090–1108. doi: 10.1002/bimj.202100190. Full text in Research Archive Show summary

  Mediation analysis in high-dimensional settings often involves identifying potential mediators among a large number of measured variables. For this purpose, a two-step familywise error rate procedure called ScreenMin has been recently proposed. In ScreenMin, variables are first screened and only those that pass the screening are tested. The proposed data-independent threshold for selection has been shown to guarantee asymptotic familywise error rate. In this work, we investigate the impact of the threshold on the finite-sample familywise error rate. We derive a power maximizing threshold and show that it is well approximated by an adaptive threshold of Wang et al. (2016, arXiv preprint arXiv:1610.03330). We illustrate the investigated procedures on a case-control study examining the effect of fish intake on the risk of colorectal adenoma. We also apply our procedure in the context of replicability analysis to identify single nucleotide polymorphisms (SNP) associated with crop yield in two distinct environments.

• Irgens, Ingebjørg; Hoff, Jana Midelfart; Jelnes, Rolf; Alexander, Marcalee; Stanghelle, Johan K & Thoresen, Magne [Show all 7 contributors for this article] (2022). Videoconferencing in Pressure Injury: Randomized Controlled Telemedicine Trial in Patients With Spinal Cord Injury. JMIR Formative Research. 6(4), p. 1–12. doi: 10.2196/27692. Full text in Research Archive Show summary

  Background: Geographical, financial and travel-related barriers may impact access to necessary health care for people in need of long-term follow-up. Objective: The goal of the research was to perform a nonblinded, randomized, controlled trial on health-related quality of life (HRQoL), healing, interaction, and satisfaction of patients with spinal cord injury (SCI) and PI receiving multidisciplinary videoconference consultations from a wound clinic to the participant’s home versus regular outpatient care. The multidisciplinary team consisted of a medical doctor, a wound nurse, and an occupational therapist. In both groups, district nurses attended the consultations at the participant’s home. Methods: A total of 56 participants, 28 in each group, were randomized to a videoconference group (VCG) or a regular care group (RCG). Validated questionnaires were used to measure and compare the follow-up effect on HRQoL. Percentage reduction of wound volume was measured at end of the follow-up. A Likert scale was used to measure the satisfaction of the patients and district nurses regarding the interaction between different modalities of care in the 2 groups. Results: The HRQoL did not show significant differences between the 2 groups (P values ranging from .09 to .88) or the rate of PI healing, experienced interaction, and satisfaction in the groups. A total of 67% (37/55) of all PIs healed, 64% (18/28) in the VCG and 70% (19/27) in the RCG. Mean reduction in ulcer volume was 79% in the VCG and 85% in the RCG (P=.32). A Kaplan-Meier plot with a logrank test regarding time to healing did not show any significant difference between the 2 groups. Conclusions: Videoconference-based care seems to be a safe and efficient way to manage PIs in terms of HRQoL, healing, interaction, and satisfaction compared to conventional care for people with SCI. This should be considered when planning for future care. SCI has a huge impact on the individual, the family, and the health care system. There is an urgent need to improve systems of care so that individuals who live far from specialists and require long-term follow-up for conditions such as PI can get optimal treatment.

• Skeie, Ivar; Clausen, Thomas; Hjemsæter, Arne Jan; Landheim, Anne Signe; Monsbakken, Bent Berntsen & Thoresen, Magne [Show all 7 contributors for this article] (2022). Mortality, Causes of Death, and Predictors of Death among Patients On and Off Opioid Agonist Treatment: Results from a 19-Year Cohort Study. European Addiction Research. ISSN 1022-6877. 28(5), p. 358–367. doi: 10.1159/000525694. Full text in Research Archive Show summary

  Background: Mortality is increased among people with opioid use disorder but reduced while on opioid agonist treatment (OAT). However, the impact of patient and treatment characteristics on mortality and causes of death is insufficiently studied. Objectives: The objective of this study was to explore mortality and causes of death and examine the impact of patient and treatment characteristics on mortality in an OAT cohort with high retention in treatment. Methods: Design: longitudinal cohort study. Setting: Norway. Observation period: time from OAT start as of 1998 until death or end of 2016, 2,508 person-years (PY) in total. Sample: 200 persons starting OAT 1998–2007. Data sources: hospital records, interviews, the Norwegian Cause of Death Registry, Statistics Norway. Results: Retention: 86.4% of the observation period was on OAT, 9.0% off, 4.6% unknown OAT status. All-cause crude mortality rate per 100 PY during the whole observation period was 1.64 (95% CI: 1.19–2.20), for deaths of somatic cause 0.88 (0.56–1.31), for drug-induced deaths 0.44 (0.23–0.76), and traumatic deaths 0.24 (0.10–0.50). Off-versus-on-OAT all-cause mortality ratio was 2.31 (1.00–4.85). On OAT, 58% of the deaths were of somatic cause and 21% drug-induced; off OAT, 38% of somatic cause and 50% drug-induced. Increasing baseline age and rate of somatic hospital treatment episodes were independently associated with increased all-cause mortality risk, while increasing rate of in-patient psychiatric treatment episodes was associated with reduced risk. Increasing duration of nicotine and cannabis use and alcohol dependence as well as increasing severity of polydrug use were associated with increased all-cause and somatic mortality adjusted for age and sex. Conclusion: The long observation period made it possible to demonstrate the importance of long-term retention in OAT to reduce mortality. Further, the preponderance of somatic and reduction of drug-induced causes of death during OAT underlines the need for follow-up of chronic diseases and health-promoting lifestyle changes. These findings add to the knowledge about long-term OAT effects, not least in ageing OAT populations. © 2022 The Author(s). This article is licensed under the Creative Commons Attribution 4.0 International License (CC BY) (http://www.karger.com/Services/OpenAccessLicense). Usage, derivative works and distribution are permitted provided that proper credit is given to the author and the original publisher.

• Christensen, Vivi Lycke; Rustøen, Tone; Thoresen, Magne; Holm, Are Martin & Bentsen, Signe Berit (2022). Stability of distinct symptom experiences in patients with chronic obstructive pulmonary disease (COPD). Respiratory Medicine. ISSN 0954-6111. 201, p. 1–8. doi: 10.1016/j.rmed.2022.106944. Full text in Research Archive Show summary

  Purpose This study aimed to examine reclassification rates among classes of chronic obstructive pulmonary disease (COPD) patients based on their distinct symptom experiences and to assess how these subgroups differed in symptom scores and health-related quality of life (HRQoL) outcomes over one year. Moreover, we wished to assess how these subgroups differed in demographic and clinical characteristics at 12 months. Patients and methods This is a follow-up study of 267 patients with moderate, severe, and very severe COPD. Based on their distinct symptom experiences using the Memorial Symptom Assessment Scale (MSAS), three subgroups (i.e., “high”, “intermediate”, and “low”) were identified at baseline. In the present study, transitions between the subgroups at three, six, nine, and 12 months were investigated and calculated as reclassification rates. Differences among the subgroups in symptom scores and HRQoL at each time point and demographic and clinical characteristics at 12 months were evaluated using analysis of variance with post hoc comparisons. Results Almost 65% were still in the “high” class after 12 months. At 12 months, pairwise comparisons for respiratory function measurements were not significantly different. Compared to the “intermediate” and “low” class, patients in the “high” class were more likely to be women and had significantly more comorbidities, reported a significantly higher number of symptoms at all time points, and worse HRQoL scores. Conclusion Our findings suggest that the pattern of a high symptom burden in COPD is consistent over time. The patients' individual symptom experiences should be the primary focus of treatment.

• Konglevoll, Dina Moxness; Hjartåker, Anette; Hopstock, Laila Arnesdatter; Strand, Bjørn Heine; Thoresen, Magne & Andersen, Lene Frost [Show all 7 contributors for this article] (2022). Protein Intake and the Risk of Pre-Frailty and Frailty in Norwegian Older Adults. The Tromsø Study 1994–2016. Journal of Frailty & Aging (JFA). ISSN 2260-1341. 11(3), p. 256–266. doi: 10.14283/jfa.2022.16. Full text in Research Archive Show summary

  Background Protein intake is suggested as an important dietary factor in the prevention of frailty, however, the influence of lifelong intake remains unclear. Objectives The present study investigated the relationship between daily protein intake and patterns of protein intake over 21 years and the risk of pre-frailty/frailty. Design Prospective cohort study. Setting The population-based Tromsø Study in Tromsø municipality, Norway. Participants In total, 1,906 women and 1,820 men aged ≥45 years in 1994 who participated in both Tromsø4 (1994–95) and Tromsø7 (2015–16). Measurements Frailty status in Tromsø7 was measured according to Fried’s phenotype, classifying participants as “robust” (frailty components present: 0), “pre-frail” (1–2) or “frail” (≥3). Daily intake of protein was estimated from self-reported habitual dietary intake using food frequency questionnaires and assessed as grams per kilogram bodyweight (g/kg BW) and per megajoule energy intake (g/MJ). The protein–frailty association was assessed via longitudinal and cross-sectional multivariable logistic regression analyses. Results The prevalence of pre-frailty and frailty in this study was 27% and 1.0%, respectively. Longitudinal analysis showed that the odds of pre-frailty/frailty decreased by 57% (odds ratio (OR) = 0.43, 95% confidence interval (CI) = 0.31;0.58, p<0.001) with the increase in intake of one additional gram of dietary protein per kg BW. The results obtained from cross-sectional analysis were similar. Tracking analysis showed that, compared to a stable high intake of protein in g/kg BW over time, other patterns of protein intake increased the risk of pre-frailty/frailty. No associations were found between intake of protein in g/MJ and pre-frailty/frailty. Conclusions Intake of protein in g/kg BW both in mid-life and later in life was inversely associated with pre-frailty/frailty in older adults. This emphasizes the importance of an adequate protein intake to facilitate healthy ageing in Norwegian older adults.

• Zhang, Chi; Eken, Torsten; Jørgensen, Silje Bakken; Thoresen, Magne & Søvik, Signe (2022). Effects of patient-level risk factors, departmental allocation and seasonality on intrahospital patient transfer patterns: network analysis applied on a Norwegian single-centre data set. BMJ Open. 12(3). doi: 10.1136/bmjopen-2021-054545. Full text in Research Archive Show summary

  Objectives Describe patient transfer patterns within a large Norwegian hospital. Identify risk factors associated with a high number of transfers. Develop methods to monitor intrahospital patient flows to support capacity management and infection control. Design Retrospective observational study of linked clinical data from electronic health records. Setting Tertiary care university hospital in the Greater Oslo Region, Norway. Participants All adult (≥18 years old) admissions to the gastroenterology, gastrointestinal surgery, neurology and orthopaedics departments at Akershus University Hospital, June 2018 to May 2019. Methods Network analysis and graph theory. Poisson regression analysis. Outcome measures Primary outcome was network characteristics at the departmental level. We describe location-to-location transfers using unweighted, undirected networks for a full-year study period. Weekly networks reveal changes in network size, density and key categories of transfers over time. Secondary outcome was transfer trajectories at the individual patient level. We describe the distribution of transfer trajectories in the cohort and associate number of transfers with patient clinical characteristics. Results The cohort comprised 17 198 hospital stays. Network analysis demonstrated marked heterogeneity across departments and throughout the year. The orthopaedics department had the largest transfer network size and density and greatest temporal variation. More transfers occurred during weekdays than weekends. Summer holiday affected transfers of different types (Emergency department-Any location/Bed ward-Bed ward/To-From Technical wards) differently. Over 75% of transferred patients followed one of 20 common intrahospital trajectories, involving one to three transfers. Higher number of intrahospital transfers was associated with emergency admission (transfer rate ratio (RR)=1.827), non-prophylactic antibiotics (RR=1.108), surgical procedure (RR=2.939) and stay in intensive care unit or high-dependency unit (RR=2.098). Additionally, gastrosurgical (RR=1.211), orthopaedic (RR=1.295) and neurological (RR=1.114) patients had higher risk of many transfers than gastroenterology patients (all effects: p<0.001). Conclusions Network and transfer chain analysis applied on patient location data revealed logistic and clinical associations highly relevant for hospital capacity management and infection control.

• Hemerik, Jesse; Thoresen, Magne & Finos, Livio (2021). Permutation testing in high-dimensional linear models: an empirical investigation. Journal of Statistical Computation and Simulation. ISSN 0094-9655. 91(5), p. 897–914. doi: 10.1080/00949655.2020.1836183. Full text in Research Archive Show summary

  Permutation testing in linear models, where the number of nuisance coefficients is smaller than the sample size, is a well-studied topic. The common approach of such tests is to permute residuals after regressing on the nuisance covariates. Permutation-based tests are valuable in particular because they can be highly robust to violations of the standard linear model, such as non-normality and heteroscedasticity. Moreover, in some cases they can be combined with existing, powerful permutation-based multiple testing methods. Here, we propose permutation tests for models where the number of nuisance coefficients exceeds the sample size. The performance of the novel tests is investigated with simulations. In a wide range of simulation scenarios our proposed permutation methods provided appropriate type I error rate control, unlike some competing tests, while having good power.

• Bøhn, Siv Kjølsrud; Myhrstad, Mari; Thoresen, Magne; Erlund, Iris; Vasstrand, Ann Kristin & Marciuch, Anne [Show all 11 contributors for this article] (2021). Bilberry/red grape juice decreases plasma biomarkers of inflammation and tissue damage in aged men with subjective memory impairment –a randomized clinical trial. BMC Nutrition. 7(75). doi: 10.1186/s40795-021-00482-8. Full text in Research Archive Show summary

  Background: Few randomized clinical trials have explored the health effects of bilberries in humans. The aim was to test the effect of bilberry and red grape-juice consumption on visual memory, motor speed and dexterity as well as inflammatory and tissue damage biomarkers of plasma in aged men with subjective memory impairment. Methods: Nine-week double-blind, placebo-controlled, dietary intervention study of aged men (n = 60, age ≥ 67 years) with subjective memory impairment randomized to consume a 50/50 mix of bilberry/red grape-juice or an iso-caloric placebo juice. A selection of Cambridge Cognition Test Battery (CANTAB), Grooved Pegboard tests and blood-sampling for biomarker analysis were performed before and after the intervention. Results: Compared to placebo the selected memory and motor test scores were un-affected by the bilberry/red grape intervention. However, the plasma levels of tissue damage biomarkers decreased significantly more in the bilberry/red grape group. In particular lactate dehydrogenase (LDH) decreased from 362 U/L (median, baseline) to 346 U/L (median, post intervention) in the bilberry/red grape group. Also, several biomarkers of inflammation (EGF, IL6, IL9, IL10 and TNFα) decreased significantly more in the bilberry/red grape group. Furthermore, several plasma polyphenols; p-coumaric acid, hippuric acid, protocatechuic acid, 3HPAA and vanillic acid, increased significantly more in the bilberry/red grape group compared to placebo with the largest increase in p-coumaric acid with 116%; from 2.2 [1.0,5.5] to 4.7 [2.8,8.1] μM/L (median [95% CL]). Conclusions: The results indicate that a nine-week bilberry/red grape juice intervention has no measurable effects on the selected memory scores in aged men experiencing memory problems but decreases the level of biomarkers of inflammation and tissue damage. Whether the dampening effects on inflammation and tissue damage biomarkers have relevance for neuroinflammatory brain pathology remains to be established.

• Shaikh, Fareeha; Kjøllesdal, Marte Karoline Råberg; Carslake, David; Thoresen, Magne & Næss, Øyvind Erik (2021). Cardiovascular risk factors in extended family members and birthweight in offspring. Journal of Developmental Origins of Health and Disease. ISSN 2040-1744. doi: 10.1017/S2040174421000076. Full text in Research Archive
• Larsen, Sunniva Veen; Holven, Kirsten B.; Christensen, Jacob J.; Flatberg, Arnar; Rundblad, Amanda & Leder, Lena [Show all 13 contributors for this article] (2021). Replacing Saturated Fat with Polyunsaturated Fat Modulates Peripheral Blood Mononuclear Cell Gene Expression and Pathways Related to Cardiovascular Disease Risk Using a Whole Transcriptome Approach. Molecular Nutrition & Food Research. ISSN 1613-4125. 65(24). doi: 10.1002/mnfr.202100633. Full text in Research Archive Show summary

  The aim of this study is to explore the molecular mechanisms underlying the effect of replacing dietary saturated fat (SFA) with polyunsaturated fat (PUFA) on cardiovascular disease (CVD) risk using a whole transcriptome approach. Methods and Results Healthy subjects with moderate hypercholesterolemia (n = 115) are randomly assigned to a control diet (C-diet) group or an experimental diet (Ex-diet) group receiving comparable food items with different fatty acid composition for 8 weeks. RNA isolated from peripheral blood mononuclear cells (PBMCs) at baseline and after 8 weeks of intervention is analyzed by microarray technology (n = 95). By use of a linear regression model (n = 92), 14 gene transcripts are differentially altered in the Ex-diet group compared to the C-diet group. These include transcripts related to vascular smooth muscle cell proliferation, low-density lipoprotein receptor folding, and regulation of blood pressure. Furthermore, pathways mainly related to immune response and inflammation, signal transduction, development, and cytoskeleton remodeling, gene expression and protein function, are differentially enriched between the groups. Conclusion Replacing dietary SFA with PUFA for 8 weeks modulates PBMC gene expression and pathways related to CVD risk in healthy subjects with moderate hypercholesterolemia.

• Stolt, Emma Theresia Kjelland; Olsen, Thomas; Elshorbagy, Amany; Kozich, Viktor; Greevenbroek, Marleen M.J. Van & Øvrebø, Bente [Show all 10 contributors for this article] (2021). Sulfur amino acid restriction, energy metabolism and obesity: a study protocol of an 8-week randomized controlled dietary intervention with whole foods and amino acid supplements. Journal of Translational Medicine. 19. doi: 10.1186/s12967-021-02824-3. Full text in Research Archive Show summary

  Background Dietary sulfur amino acid (SAA) restriction is an established animal model for increasing lifespan and improving metabolic health. Data from human studies are limited. In the study outlined in this protocol, we will evaluate if dietary SAA restriction can reduce body weight and improve resting energy expenditure (REE) and parameters related to metabolic health. Method/design Men and women (calculated sample size = 60), aged 18–45 years, with body mass index of 27–35 kg/m2 will be included in a double-blind 8-week dietary intervention study. The participants will be randomized in a 1:1 manner to a diet with either low or high SAA. Both groups will receive an equal base diet consisting of low-SAA plant-based whole foods and an amino acid supplement free of SAA. Contrasting SAA contents will be achieved using capsules with or without methionine and cysteine (SAAhigh, total diet SAA ~ 50–60 mg/kg body weight/day; SAAlow, total diet SAA ~ 15–25 mg/kg body weight/day). The primary outcome is body weight change. Data and material collection will also include body composition (dual X-ray absorptiometry), resting energy expenditure (whole-room indirect calorimetry) and samples of blood, urine, feces and adipose tissue at baseline, at 4 weeks and at study completion. Measures will be taken to promote and monitor diet adherence. Data will be analyzed using linear mixed model regression to account for the repeated measures design and within-subject correlation. Discussion The strength of this study is the randomized double-blind design. A limitation is the restrictive nature of the diet which may lead to poor compliance. If this study reveals a beneficial effect of the SAAlow diet on body composition and metabolic health, it opens up for new strategies for prevention and treatment of overweight, obesity and its associated disorders. Trial registration ClinicalTrials.gov: NCT04701346, Registration date: January 8th, 2021

• Ghosh, Abhik & Thoresen, Magne (2021). Consistent fixed-effects selection in ultra-high dimensional linear mixed models with error-covariate endogeneity. Statistica sinica. ISSN 1017-0405. 31(3), p. 2073–2102. doi: 10.5705/ss.202019.0421. Full text in Research Archive
• Ghosh, Abhik & Thoresen, Magne (2021). A robust variable screening procedure for ultra-high dimensional data. Statistical Methods in Medical Research. ISSN 0962-2802. 30(8), p. 1816–1832. doi: 10.1177/09622802211017299. Full text in Research Archive Show summary

  Variable selection in ultra-high dimensional regression problems has become an important issue. In such situations, penalized regression models may face computational problems and some pre-screening of the variables may be necessary. A number of procedures for such pre-screening has been developed; among them the Sure Independence Screening (SIS) enjoys some popularity. However, SIS is vulnerable to outliers in the data, and in particular in small samples this may lead to faulty inference. In this paper, we develop a new robust screening procedure. We build on the density power divergence (DPD) estimation approach and introduce DPD-SIS and its extension iterative DPD-SIS. We illustrate the behavior of the methods through extensive simulation studies and show that they are superior to both the original SIS and other robust methods when there are outliers in the data. Finally, we illustrate its use in a study on regulation of lipid metabolism.

• Ponzi, Erica; Thoresen, Magne; Nøst, Therese Haugdahl & Møllersen, Kajsa (2021). Integrative, multi-omics, analysis of blood samples improves model predictions: applications to cancer. BMC Bioinformatics. 22(1), p. 1–17. doi: 10.1186/s12859-021-04296-0. Full text in Research Archive Show summary

  <i>Background</i> - Cancer genomic studies often include data collected from several omics platforms. Each omics data source contributes to the understanding of the underlying biological process via source specific (“individual”) patterns of variability. At the same time, statistical associations and potential interactions among the different data sources can reveal signals from common biological processes that might not be identified by single source analyses. These common patterns of variability are referred to as “shared” or “joint”. In this work, we show how the use of joint and individual components can lead to better predictive models, and to a deeper understanding of the biological process at hand. We identify joint and individual contributions of DNA methylation, miRNA and mRNA expression collected from blood samples in a lung cancer case–control study nested within the Norwegian Women and Cancer (NOWAC) cohort study, and we use such components to build prediction models for case–control and metastatic status. To assess the quality of predictions, we compare models based on simultaneous, integrative analysis of multi-source omics data to a standard non-integrative analysis of each single omics dataset, and to penalized regression models. Additionally, we apply the proposed approach to a breast cancer dataset from The Cancer Genome Atlas.<br><br> <i>Results</i> - Our results show how an integrative analysis that preserves both components of variation is more appropriate than standard multi-omics analyses that are not based on such a distinction. Both joint and individual components are shown to contribute to a better quality of model predictions, and facilitate the interpretation of the underlying biological processes in lung cancer development.<br><br> <i>Conclusions</i> - In the presence of multiple omics data sources, we recommend the use of data integration techniques that preserve the joint and individual components across the omics sources. We show how the inclusion of such components increases the quality of model predictions of clinical outcomes.

• Zhang, Chi; Fanaee-T, Hadi & Thoresen, Magne (2021). Feature extraction from unequal length heterogeneous EHR time series via dynamic time warping and tensor decomposition. Data mining and knowledge discovery. ISSN 1384-5810. 35, p. 1760–1784. doi: 10.1007/s10618-020-00724-6. Full text in Research Archive
• Christensen, Jacob Juel; Ulven, Stine Marie; Thoresen, Magne; Westerman, Kenneth; Holven, Kirsten Bjørklund & Andersen, Lene Frost (2020). Associations between dietary patterns and gene expression pattern in peripheral blood mononuclear cells: a cross-sectional study (preprint). BioRxiv. ISSN 0362-4331. doi: 10.1101/2020.01.25.20018465. Full text in Research Archive
• Rosseland, Leiv Arne; Reme, Silje Endresen; Simonsen, Tone Breines; Thoresen, Magne; Nielsen, Christopher Sivert & Eberhard-Gran, Malin (2020). Are labor pain and birth experience associated with persistent pain and postpartum depression? A prospective cohort study. Scandinavian Journal of Pain. ISSN 1877-8860. 20(3), p. 591–602. doi: 10.1515/sjpain-2020-0025. Full text in Research Archive Show summary

  <i>Background and aims</i> - A considerable research-literature focuses on pain during labor and associations with postpartum persistent pain and depression, with findings pointing in various directions. The aim of this study was to examine the role of labor pain and overall birth experience in the development of pain and depression 8 weeks after delivery.<p><p> <i>Methods</i> - The study sample was drawn from the Akershus Birth Cohort. Data from multiple sources were used, including the hospital’s birth record (n = 4,391), questionnaire data from gestational week 17 of pregnancy (n = 3,752), 8 weeks postpartum (n = 2,217), and two questions about pain and birth experience asked within 48 h after delivery (n = 1,221). The Edinburgh Postnatal Depression Scale was used to measure postpartum depression, a single question was used to measure persistent pain 8 weeks postpartum, while pain and birth experience were measured by numeric rating scales. A history of pre-pregnant depression and chronic pain were measured through self-report questions in gestational week 17. A total of 645 women had complete data from all sources. We applied multiple imputation techniques to handle missing responses on the two questions about pain and birth experience.<p><p> <i>Results</i> - The results showed that neither labor pain nor birth experience were associated with persistent pain 8 weeks postpartum, whereas pain before pregnancy (OR 3.70; 95% CI 2.71–5.04) and a history of depression (OR 2.31; 95% CI 1.85–2.88) were statistically significant predictors of persistent pain. A negative birth experience was significantly (OR 1.16; 95% CI 1.04–1.29) associated with postpartum depression, whereas labor pain intensity was not. A history of depression (OR 3.95; 95% CI 2.92–5.34) and pre-pregnancy pain (OR 2.03; 95% CI 1.37–3.01) were important predictors of postpartum depression 8 weeks after delivery.<p><p> <i>Conclusions and implications</i> - Whilst the relationship between labor pain intensity and postpartum pain and depression remain unclear, our results do imply the need to screen for previous depression and chronic pain conditions in pregnant women, as well as consider preventive measures in those who screen positive.

• Christensen, Jacob Juel; Ulven, Stine Marie; Thoresen, Magne; Westerman, Kenneth; Holven, Kirsten Bjørklund & Andersen, Lene Frost (2020). Associations between dietary patterns and gene expression pattern in peripheral blood mononuclear cells: A cross-sectional study. NMCD. Nutrition Metabolism and Cardiovascular Diseases. ISSN 0939-4753. 30(11), p. 2111–2122. doi: 10.1016/j.numecd.2020.06.018. Full text in Research Archive Show summary

  Background and aims Diet may alter gene expression in immune cells involved in atherosclerotic cardiovascular disease susceptibility. However, we still lack a robust understanding of the association between diet and immune cell-related gene expression in humans. Therefore, we examined associations between dietary patterns (DPs) and gene expression profiles in peripheral blood mononuclear cells (PBMCs) in a population of healthy, Norwegian adults (n = 130 women and 105 men). Methods and results We used factor analysis to define a posteriori DPs from food frequency questionnaire-based dietary assessment data. In addition, we derived interpretable features from microarray-based gene expression data (13 967 transcripts) using two algorithms: CIBERSORT for estimation of cell subtype proportions, and weighted gene co-expression network analysis (WGCNA) for cluster discovery. Finally, we associated DPs with either CIBERSORT-predicted PBMC leukocyte distribution or WGCNA gene clusters using linear regression models. We detected three DPs that broadly reflected Western, Vegetarian, and Low carbohydrate diets. CIBERSORT-predicted percentage of monocytes associated negatively with the Vegetarian DP. For women, the Vegetarian DP associated with a large gene cluster consisting of 600 genes mainly involved in regulation of DNA transcription, whereas for men, the Western DP inversely associated with a smaller cluster of 36 genes mainly involved in regulation of metabolic and inflammatory processes. A subsequent protein–protein interaction network analysis suggested that genes within these clusters might physically interact in biological networks. Conclusions Although the present findings are exploratory, our analysis pipeline serves as a useful framework for studying the association between diet and gene expression.

• Page, Christian; Djordjilovic, Vera; Nøst, Therese Haugdahl; Ghiasvand, Reza; Sandanger, Torkjel M & Rattalma, Arnoldo Frigessi Di [Show all 8 contributors for this article] (2020). Lifetime ultraviolet radiation exposure and DNA methylation in blood leukocytes: The Norwegian Women and Cancer Study. Scientific Reports. 10, p. 1–8. doi: 10.1038/s41598-020-61430-3. Full text in Research Archive Show summary

  Ultraviolet radiation (UVR) exposure is a leading cause of skin cancers and an ubiquitous environmental exposure. However, the molecular mechanisms relating UVR exposure to melanoma is not fully understood. We aimed to investigate if lifetime UVR exposure could be robustly associated to DNA methylation (DNAm). We assessed DNAm in whole blood in three data sets (n = 183, 191, and 125) from the Norwegian Woman and Cancer cohort, using Illumina platforms. We studied genome-wide DNAm, targeted analyses of CpG sites indicated in the literature, global methylation, and accelerated aging. Lifetime history of UVR exposure (residential ambient UVR, sunburns, sunbathing vacations and indoor tanning) was collected by questionnaires. We used one data set for discovery and the other two for replication. One CpG site showed a genome-wide significant association to cumulative UVR exposure (cg01884057) (pnominal = 3.96e-08), but was not replicated in any of the two replication sets (pnominal ≥ 0.42). Two CpG sites (cg05860019, cg00033666) showed suggestive associations with the other UVR exposures. We performed extensive analyses of the association between long-term UVR exposure and DNAm. There was no indication of a robust effect of past UVR exposure on DNAm.

• Fanaee-T, Hadi & Thoresen, Magne (2020). Iterative Multi-mode Discretization: Applications to Co-clustering. Lecture Notes in Computer Science (LNCS). ISSN 0302-9743. 12323, p. 94–105. doi: 10.1007/978-3-030-61527-7_7. Full text in Research Archive Show summary

  We introduce a new concept called “Iterative Multi-Mode Discretization (IMMD)” which is a new type of efficient data sparsification that can scale up many tasks in data mining. In this paper we demonstrate the application of IMMD in co-clustering, i.e. simultaneous clustering of the rows and columns in a matrix. We propose IMMD-CC, a novel co-clustering algorithm, which is developed based on IMMD. IMMD-CC has attractive properties. First, its time complexity is linear, so it can be used in large-scale problems. In addition, IMMD-CC is able to estimate the number of co-clusters automatically, and more accurate than state-of-the-art methods. We demonstrate the performance of IMMD-CC in comparison to several state-of-the-art methods on 100 data sets from a benchmark cohort, as well as 35 real-world datasets. The results show the promising potential of the proposed method.

• Kollerud, Ruby Del Risco; Haugnes, Hege Sagstuen; Claussen, Bjørgulf; Thoresen, Magne; Nafstad, Per & Farnham, James [Show all 9 contributors for this article] (2020). A population-based study of testicular cancer risk among children and young adults from Norway and Utah, USA. International Journal of Cancer. ISSN 0020-7136. 147(6). doi: 10.1002/ijc.32949. Full text in Research Archive Show summary

  Similar family-based cancer and genealogy data from Norway and Utah allowed comparisons of the incidence of testicular cancer (TC), and exploration of the role of Scandinavian ancestry and family history of TC in TC risk. Our study utilizes data from the Utah Population Database and Norwegian Population Registers. All males born during 1951–2015 were followed for TC until the age of 29 years. A total of 1,974,287 and 832,836 males were born in Norway and Utah, respectively, of whom 2,686 individuals were diagnosed with TC in Norway and 531 in Utah. The incidence per year of TC in Norway (10.6) was twice that observed in Utah (5.1) for males born in the last period (1980–1984). The incidence rates of TC in Utah did not differ according to the presence or absence of Scandinavian ancestry (p = 0.669). Having a brother diagnosed with TC was a strong risk factor for TC among children born in Norway and Utah, with HR = 9.87 (95% CI 5.68–17.16) and 6.02 (95% CI 4.80–7.55), respectively; with even higher HR observed among the subset of children in Utah with Scandinavian ancestry (HR = 12.30, 95% CI 6.78–22.31). A clear difference in TC incidence among individuals born in Norway and descendants of Scandinavian people born in Utah was observed. These differences in TC rates point to the possibility of environmental influence. Family history of TC is a strong risk factor for developing TC in both populations.

• Rundblad, Amanda; Holven, Kirsten Bjørklund; Øyri, Linn Kristin Lie; Hansson, Patrik; Ivan, Ingvild Hansen & Gjevestad, Gyrd Omholt [Show all 8 contributors for this article] (2020). Intake of fermented dairy products induces a less pro-inflammatory postprandial peripheral blood mononuclear cell gene expression response than non-fermented dairy products: A randomized controlled cross-over trial. Molecular Nutrition & Food Research. ISSN 1613-4125. 64(21), p. 1–11. doi: 10.1002/mnfr.202000319. Full text in Research Archive Show summary

  Scope: It is aimed to investigate how intake of high-fat meals composed of different dairy products with a similar fat content affects postprandial peripheral blood mononuclear cell (PBMC) expression of inflammation-related genes, as well as circulating inflammatory markers and metabolites.<p>Methods and results: Healthy subjects (n = 47) consume four different high-fat meals composed of either butter, cheese, whipped cream, or sourcream in a randomized controlled cross-over study. Fasting and postprandial PBMC gene expression, plasma metabolites, and circulating inflammatory markers are measured. Using a linear mixed model, it is found that expression of genes related to lymphocyte activation, cytokine signaling, chemokine signaling, and cell adhesion is differentially altered between the four meals. In general, intake of the fermented products cheese and sour cream reduces,while intake of the non-fermented products butter and whipped cream increases, expression of these genes. Plasma amino acid concentrations increase after intake of cheese compared to the other meals, and the aminoacid changes correlate with several of the differentially altered genes.<p>Conclusion: Intake of fermented dairy products, especially cheese, induces a less inflammatory postprandial PBMC gene expression response than non-fermented dairy products. These findings may partly explain inconsistent findings in studies on health effects of dairy products.

• Irgens, Ingebjørg; Hoff, Jana Midelfart; Jelnes, Rolf; Alexander, Marcalee; Stanghelle, Johan K & Thoresen, Magne [Show all 7 contributors for this article] (2020). Spinal cord injury and development of pressure injury during acute rehabilitation in Norway: a national retrospective cross-sectional study. Spinal Cord. ISSN 1362-4393. 58, p. 1069–1079. doi: 10.1038/s41393-020-0465-z. Full text in Research Archive Show summary

  Study design A national, retrospective, cross-sectional study. Objectives To analyze the prevalence of pressure injury (PI), and characteristics associated with PI development in the hospitalized population of persons with a newly acquired spinal cord injury (SCI) between 2004 and 2014. Setting All three specialized Spinal Cord Units in Norway. Methods Demographic data related to prevalence and potential risk factors were retrieved from the electronic medical record (EMR). Statistical analyses were performed, using IBM SPSS Statistics, version 23. Results We identified 1012 individuals with a new SCI. Mean age at injury was 48 years (SD 19). The period prevalence of PI was 16% (95% CI = 0.14–0.19), and identified PI associations were complete SCI (OR = 0.1), being injured abroad (OR = 2.4), bowel (OR = 13), and bladder (OR = 9.2) dysfunction; comorbidities like diabetes mellitus 1 (OR = 7.9), diagnosed depression (OR = 3.8), ventilator support (OR = 3.0), drug abuse (OR = 3.0), and concurrent traumatic brain injury (OR = 1.7). Individuals in the age group of 15–29 years had higher odds of PI compared with middle-aged individuals (45–59 years). Conclusion PI is a serious complication after SCI. The association between depression or comorbidity and PI occurrence should be investigated more thoroughly. We recommend implementation of a simple follow-up program regarding observation and prevention of PI. Increased awareness of factors that could contribute to PI will help to focus on better prevention and early recognition of PI. This will contribute to more optimal rehabilitation.

• Kjeverud, Anita; Østlie, Kristin; Schanke, Anne-Kristine; Gay, Caryl; Thoresen, Magne & Lerdal, Anners (2020). Trajectories of fatigue among stroke patients from the acute phase to 18 months post-injury: A latent class analysis. PLOS ONE. 15(4), p. 1–14. doi: 10.1371/journal.pone.0231709. Full text in Research Archive Show summary

  Using GMM, three latent classes of fatigue trajectories over 18 months were identified, characterized by differing levels of fatigue: low, moderate and high. The mean FSS score for each class remained relatively stable across all four time points. In the univariate analyses, age

• Kollerud, Ruby Del Risco; Cannon-Albright, LA; Haugnes, Hege Sagstuen; Ruud, Ellen; Thoresen, Magne & Nafstad, Per [Show all 9 contributors for this article] (2020). Childhood central nervous system tumors and leukemia: Incidence and familial risk. A comparative population-based study in Utah and Norway. Pediatric Blood & Cancer. ISSN 1545-5009. Aug;67(8):e28408. doi: 10.1002/pbc.28408. Full text in Research Archive Show summary

  Background In this study, we aimed to evaluate incidence rates and family risk of the most common childhood cancers, tumors in the central nervous system (CNS), and leukemia among individuals from Norway and individuals with Scandinavian ancestry living in Utah. Methods We used the Utah Population Database and the Norwegian National Population Register linked to Cancer registries to identify cancers in children born between 1966 and 2015 and their first‐degree relatives. We calculated incidence rates and hazards ratios. Results The overall incidence of CNS tumors increased with consecutive birth cohorts similarly in Utah and Norway (both P < 0.001). Incidence rates of leukemia were more stable and similar in both Utah and in Norway with 4.6/100 000 person‐years among children (<15 years) born in the last cohort. A family history of CNS tumors was significantly associated with risk of childhood CNS tumors in Utah HR = 3.05 (95% CI 1.80‐5.16) and Norway HR = 2.87 (95% CI 2.20‐3.74). In Norway, children with a first‐degree relative diagnosed with leukemia had high risk of leukemia (HR = 2.39, 95% CI 1.61‐3.55). Conclusion Despite geographical distance and assumed large lifestyle differences, two genetically linked pediatric populations show similar incidences of CNS tumors and leukemia in the period 1966–2015. CNS tumors and leukemia aggregated in families in both countries.

• Hjemsæter, Arne Jan; Bramness, Jørgen Gustav; Drake, Robert; Skeie, Ivar; Monsbakken, Bent Berntsen & Thoresen, Magne [Show all 7 contributors for this article] (2019). Predictors of problematic substance use 18 years after treatment: a longitudinal cohort study of persons with substance use disorders. Cogent Psychology. 6:1634325, p. 1–13. doi: 10.1080/23311908.2019.1634325. Full text in Research Archive Show summary

  Objective: To examine the extent of substance use and explore which baseline factors predicted current problematic substance use 18 years after treatment in surviving patients. Methods: This longitudinal cohort study used a mailed self-report questionnaire on a group of patients with long-term problematic substance use, and high psychiatric comorbidity, 18 years after they received treatment for substance use disorders. A consecutive sample of patients with substance use disorders (n = 287, mean age 38.6 years, 72% male; 45% alcohol use disorder only and 55% other substance use disorders) was recruited in 1997 and 1998. Baseline measurements included the Composite International Diagnostic Interview, the Millon Clinical Multiaxial Inventory II, the Hopkins Symptom Checklist 25 and demographic data. In the 18-year follow-up study (n = 91), the Alcohol Use Disorders Identification Test and the Drug Use Disorders Identification Test were used. A multivariate logistic regression analysis was used to identify predictors of having current problematic substance use 18 years after treatment for participants with full data collection at baseline (n = 75). Results: Of the 91 participants who responded at the 18-year follow-up study, 47 (52%) had no current problematic substance use and 44 (48%) had current problematic substance use during the past 12 months. Of participants with no current problematic substance use, 17 (36%) reported total abstinence from alcohol, illicit substances and misuse of pharmaceuticals during the past 5 years. Mental distress measured with the Hopkins Symptom Checklist 25 at baseline predicted having current problematic substance use 18 years after treatment (n = 75). Conclusions: In a group of patients with long lasting substance use disorders and high psychiatric comorbidity that entered treatment, close to half of those participating at a follow-up study had current problematic substance use during the past 12 months 18 years later. Mental distress at baseline predicted having current problematic substance use 18 years after treatment, adjusted for other factors. Self-reported general mental health symptoms can be a relevant predictor of the long-term course of substance use disorders.

• Tork, Hadi Fanaee & Thoresen, Magne (2019). Multi-insight visualization of multi-omics data via ensemble dimension reduction and tensor factorization. Bioinformatics. ISSN 1367-4803. 35(10), p. 1625–1633. doi: 10.1093/bioinformatics/bty847. Full text in Research Archive
• Hansson, Patrik; Holven, Kirsten Bjørklund; Øyri, Linn Kristin Lie; Brekke, Hilde Kristin; Gjevestad, Gyrd Omholt & Thoresen, Magne [Show all 7 contributors for this article] (2019). Sex differences in postprandial responses to different dairy products on lipoprotein subclasses: A randomised controlled cross-over trial. British Journal of Nutrition. ISSN 0007-1145. 122(7), p. 780–789. doi: 10.1017/S0007114519001429. Full text in Research Archive Show summary

  Men have earlier first-time event of CHD and higher postprandial TAG response compared with women. The aim of this exploratory sub-study was to investigate if intake of meals with the same amount of fat from different dairy products affects postprandial lipoprotein subclasses differently in healthy women and men. A total of thirty-three women and fourteen men were recruited to a randomised controlled cross-over study with four dairy meals consisting of butter, cheese, whipped cream or sour cream, corresponding to 45 g of fat (approximately 60 energy percent). Blood samples were taken at 0, 2, 4 and 6 h postprandially. Lipoprotein subclasses were measured using NMR and analysed using a linear mixed model. Sex had a significant impact on the response in M-VLDL (P=0·04), S-LDL (P=0·05), XL-HDL (P=0·009) and L-HDL (P=0·001) particle concentration (P), with women having an overall smaller increase in M-VLDL-P, a larger decrease in S-LDL-P and a larger increase in XL- and L-HDL-P compared with men, independent of meal. Men showed a decrease in XS-VLDL-P compared with women after intake of sour cream (P&amp;lt;0·01). In men only, XS-VLDL-P decreased after intake of sour cream compared with all other meals (v. butter: P=0·001; v. cheese: P=0·04; v. whipped cream: P=0·006). Meals with the same amount of fat from different dairy products induce different postprandial effects on lipoprotein subclass concentrations in men and women.

• Djordjilovic, Vera; Page, Christian Magnus; Gran, Jon Michael; Nøst, Therese Haugdahl; Sandanger, Torkjel M & Veierød, Marit Bragelien [Show all 7 contributors for this article] (2019). Global test for high-dimensional mediation: Testing groups of potential mediators. Statistics in Medicine. ISSN 0277-6715. 38(18), p. 3346–3360. doi: 10.1002/sim.8199. Full text in Research Archive Show summary

  We address the problem of testing whether a possibly high‐dimensional vector may act as a mediator between some exposure variable and the outcome of interest. We propose a global test for mediation, which combines a global test with the intersection‐union principle. We discuss theoretical properties of our approach and conduct simulation studies that demonstrate that it performs equally well or better than its competitor. We also propose a multiple testing procedure, ScreenMin, that provides asymptotic control of either familywise error rate or false discovery rate when multiple groups of potential mediators are tested simultaneously. We apply our approach to data from a large Norwegian cohort study, where we look at the hypothesis that smoking increases the risk of lung cancer by modifying the level of DNA methylation.

• Rundblad, Amanda; Larsen, Sunniva; Myhrstad, Mari; Ottestad, Inger; Thoresen, Magne & Holven, Kirsten Bjørklund [Show all 7 contributors for this article] (2019). Differences in peripheral blood mononuclear cell gene expression and triglyceride composition in lipoprotein subclasses in plasma triglyceride responders and non-responders to omega-3 supplementation. Genes & Nutrition. ISSN 1555-8932. 14(10), p. 1–13. doi: 10.1186/s12263-019-0633-y. Full text in Research Archive Show summary

  Background:Intake of the marine omega-3 fatty acids eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA)reduces fasting triglyceride (TG) levels and may thereby lower cardiovascular disease risk. However, there are largeinter-individual differences in the TG-lowering effect of omega-3 supplementation. Genotype differences partly explainthis variation, but gene-environment interactions leading to gene expression differences may also be important. In thisstudy, we aimed to investigate baseline differences and differences in the change in peripheral blood mononuclear cell(PBMC) gene expression and lipoprotein subclass TG levels between TG responders and non-responders to omega-3fatty acid supplementation.Methods:In a previous randomized controlled trial, healthy normotriglyceridemic subjects (n= 35, 71% women)received 1.6 g EPA + DHA/day for 7 weeks. In this exploratory sub-study, we defined TG responders as subjects havinga TG reduction beyond the 20% day-to-day variation and non-responders as having a TG change between−20% and+ 20% after omega-3 supplementation. PBMC gene expression was measured using microarray, and lipoproteinsubclasses were measured using nuclear magnetic resonance spectroscopy.Results:Eight subjects were defined as responders with a median TG reduction of 37%, and 16 subjects were definedas non-responders with a median TG change of 0%. At baseline, responders had higher TG levels in two of four high-density lipoprotein (HDL) subclasses and 909 gene transcripts (p≤0.05) were differentially expressed compared to non-responders. During the intervention, the plasma TG reduction among responders was reflected in TG reductions in fourof six different very low-density lipoprotein subclasses and three of four different HDL subclasses. Compared to non-responders, the expression of 454 transcripts was differentially altered in responders (p≤0.05). Pathway analyses revealed that responders had altered signaling pathways related to development and immune function. In addition,two of the top 10 enriched pathways in responders compared to non-responders were related to lysophosphatidic acid signaling.Conclusion:TG responders and non-responders to omega-3 supplementation have different lipoprotein subclass andPBMC gene expression profiles at baseline and different lipoprotein subclass and PBMC gene expression responses toomega-3 supplementation. These gene expression differences may partially explain the variability in TG response observed after omega-3 supplementation.

• Øyri, Linn Kristin Lie; Narverud, Ingunn; Bogsrud, Martin Prøven; Hansson, Patrik; Leder, Lena & Byfuglien, Marte Gjeitung [Show all 10 contributors for this article] (2019). Postprandial changes in gene expression of cholesterol influx and efflux mediators after intake of SFA compared to n-6 PUFA in subjects with and without familial hypercholesterolemia – secondary outcomes of a randomized controlled trial. Journal of Nutritional Science (JNS). 8:e27, p. 1–27. doi: 10.1017/jns.2019.25. Full text in Research Archive Show summary

  The long-term cholesterol lowering effect of replacing intake of saturated fatty acids (SFAs) with polyunsaturated fatty acids (PUFAs) is well established, however not fully explained mechanistically. We examined the postprandial response of meals with different fat quality on expression of lipid genes in peripheral blood mononuclear cells (PBMC) in subjects with and without familial hypercholesterolemia (FH). Thirteen subjects with FH (who had discontinued lipid-lowering treatment ≥4 weeks prior to both test days) and 14 normolipidemic controls were included in a randomized controlled double-blind crossover study with two meals, each with 60 grams of fat either mainly SFAs (~40 E%) or n-6 PUFAs (~40 E%). PBMCs were isolated in fasting, and 4 and 6 hours postprandial blood samples. Expression of 33 lipid genes was analysed by reverse transcription quantitative polymerase chain reaction. A linear mixed model was used to assess postprandial effects between meals and groups. There was a significant interaction between meal and group for MSR1 (p=0.03), where intake of SFAs compared to n-6 PUFAs induced a larger reduction in gene expression in controls only (p=0.01). Intake of SFAs compared to n-6 PUFAs induced larger reduction in gene expression levels of LDLR and FADS1/2, smaller increase of INSIG1 and FASN, and larger increase of ABCA1 and ABCG1 (p=0.01 for all, no group interaction). Intake of SFAs compared to n-6 PUFAs induced changes in gene expression of cholesterol influx and efflux mediators in PBMCs including lower LDLR and higher ABCA1/G1, potentially explaining the long-term cholesterol-raising effect of a high SFA intake.

• Kollerud, Ruby Del Risco; Ruud, Ellen; Haugnes, Hege Sagstuen; Cannon-Albright, Lisa A.; Thoresen, Magne & Nafstad, Per [Show all 10 contributors for this article] (2019). Family history of cancer and risk of paediatric and young adult’s testicular cancer: A Norwegian cohort study. British Journal of Cancer. ISSN 0007-0920. 120, p. 1007–1014. doi: 10.1038/s41416-019-0445-2. Full text in Research Archive Show summary

  <i>Background</i> - The aim of this study was to examine the association of a family history of cancer with the risk of testicular cancer in young adults.<p><p> <i>Methods</i> - This is a prospective cohort study including 1,974,287 males born 1951–2015, of whom 2686 were diagnosed with TC before the age of 30.<p><p> <i>Results</i> - A history of TC in male relatives was significantly associated with a diagnosis of TC among children and young adults, including brothers (6.3-fold), sons (4.7-fold), fathers (4.4-fold), paternal uncles (2.0-fold) and maternal uncles (1.9-fold). Individuals with a father diagnosed with a carcinoma or sarcoma showed an elevated risk (1.1-fold and 1.8-fold, respectively). A family history of mesothelioma was positively associated with a risk of TC [(father (2.8-fold), mother (4.6-fold) and maternal uncles and aunt (4.4-fold)]. Elevated risks were also observed when siblings were diagnosed with malignant melanoma (1.4-fold). The risk of TC was also increased when fathers (11.1-fold), paternal (4.9-fold) and maternal uncles and aunts (4.6-fold) were diagnosed with malignant neuroepithelial-tumours.<p><p> <i>Conclusion</i> - We found an increased risk of TC among children and young adults with a family history of TC, carcinoma, mesothelioma, sarcoma, malignant melanoma and malignant neuroepithelial tumours. Hereditary cancer syndromes might underlie some of the associations reported in this study.

• Myhrvold, Birgitte Lawaetz; Kongsted, Alice; Irgens, Pernille Marie; Robinson, Hilde Stendal; Thoresen, Magne & Vøllestad, Nina Køpke (2019). Broad External Validation and Update of a Prediction Model for Persistent Neck Pain After 12 Weeks. Spine. ISSN 0362-2436. 44(22), p. E1298–E1310. doi: 10.1097/BRS.0000000000003144. Full text in Research Archive Show summary

  Study Design. A prospective observational study. Objective. To externally validate the prediction model developed by Schellingerhout and colleagues predicting global perceived effect at 12 weeks in patients with neck pain and to update and internally validate the updated model. Summary of Background Data. Only one prediction model for neck pain has undergone some external validation with good promise. However, the model needs testing in other populations before implementation in clinical practice. Methods. Patients with neck pain (n = 773) consulting Norwegian chiropractors were followed for 12 weeks. Parameters from the original prediction model were applied to this sample for external validation. Subsequently, two random samples were drawn from the full study sample. One sample (n = 436) was used to update the model; by recalibration, removing noninformative covariates, and adding new possible predictors. The updated model was tested in the other sample (n = 303) using stepwise logistic regression analysis. Main outcomes for performance of models were discrimination and calibration plots. Results. Three hundred seventy patients (47%) in the full study sample reported persistent pain at 12 weeks. The performance of the original model was poor, area under the receiver operating characteristics curve was 0.55 with a Confidence Interval of 0.51–0.59. The updated model included Radiating pain to shoulder and/or elbow, education level, physical activity, consultation-type (first- time, follow-up or maintenance consultation), expected course of neck pain, previous course of neck pain, number of pain sites, and the interaction term Physical activity##Number of pain sites. The area under the receiver operating characteristics curve was 0.65 with a 95% Confidence Interval of 0.58–0.71 for the updated model. Conclusion. The predictive accuracy of the original model performed insufficiently in the sample of patients from Norwegian chiropractors and the model is therefore not recommended for that setting. Only one predictor from the original model was retained in the updated model, which demonstrated reasonable good performance predicting outcome at 12 weeks. Before considering clinical use, a new external validation is required. Level of Evidence: 3

• Thoresen, Magne (2019). Spurious interaction as a result of categorization 01 Mathematical Sciences 0104 Statistics. BMC Medical Research Methodology. 19(1), p. 1–8. doi: 10.1186/s12874-019-0667-2. Full text in Research Archive
• Romeo, Giovanni & Thoresen, Magne (2019). Model selection in high-dimensional noisy data: a simulation study. Journal of Statistical Computation and Simulation. ISSN 0094-9655. 89(11), p. 2031–2050. doi: 10.1080/00949655.2019.1607345. Full text in Research Archive Show summary

  In many practical applications, high-dimensional regression analyses have to take into account measurement error in the covariates. It is thus necessary to extend regularization methods, that can handle the situation where the number of covariates p largely exceed the sample size n, to the case in which covariates are also mismeasured. A variety of methods are available in this context, but many of them rely on knowledge about the measurement error and the structure of its covariance matrix. In this paper, we set the goal to compare some of these methods, focusing on situations relevant for practical applications. In particular, we will evaluate these methods in setups in which the measurement error distribution and dependence structure are not known and have to be estimated from data. Our focus is on variable selection, and the evaluation is based on extensive simulations.

• Hansson, Patrik; Holven, Kirsten Bjørklund; Øyri, Linn Kristin Lie; Brekke, Hilde Kristin; Biong, Anne Sofie & Gjevestad, Gyrd Omholt [Show all 10 contributors for this article] (2019). Meals with Similar Fat Content from Different Dairy Products Induce Different Postprandial Triglyceride Responses in Healthy Adults: A Randomized Controlled Cross-Over Trial. Journal of Nutrition. ISSN 0022-3166. 149(3), p. 422–431. doi: 10.1093/jn/nxy291. Full text in Research Archive Show summary

  Postprandial lipemia is a risk factor for cardiovascular disease. Dairy products differ in nutrient content and food matrix, and little is known about how different dairy products affect postprandial triglyceride (TG) concentrations. We investigated the effect of meals with similar amounts of fat from different dairy products on postprandial TG concentrations over 6 h in healthy adults. A randomized controlled cross-over study was performed on 47 subjects (30% men), with median (25th–75th percentile) age of 32 (25–46) y and body mass index of 23.6 (21.0–25.8) kg/m2. Meals included 1 of butter, cheese, whipped cream, or sour cream, corresponding to 45 g of fat (approximately 60 energy%). Serum concentrations of TGs (primary outcome), and total cholesterol (TC), low density lipoprotein cholesterol (LDL cholesterol), high density lipoprotein cholesterol (HDL cholesterol), insulin, glucose, non-esterified fatty acids, and plasma glucose-dependent insulinotropic polypeptide (secondary outcomes) were measured before the meal and 2, 4, and 6 h postprandially. Incremental AUC (iAUC) was calculated for the responses, and data were analyzed using a linear mixed model. Sour cream induced a 61% larger TG-iAUC0–6 h compared to whipped cream (P &lt; 0.001), a 53% larger TG-iAUC0–6 h compared to butter (P &lt; 0.001), and a 23% larger TG-iAUC0–6 h compared to cheese (P = 0.05). No differences in TG-iAUC0–6 h between the other meals were observed. Intake of sour cream induced a larger HDL cholesterol-iAUC0–6 h compared to cheese (P = 0.01). Intake of cheese induced a 124% larger insulin iAUC0–6 h compared to butter (P = 0.006). No other meal effects were observed. High-fat meals containing similar amount of fat from different dairy products induce different postprandial effects on serum TGs, HDL cholesterol, and insulin in healthy adults. The potential mechanisms and clinical impact of our findings remain to be further elucidated. The study was registered at www.clinicaltrials.gov as NCT02836106.

• Tork, Hadi Fanaee & Thoresen, Magne (2019). Performance evaluation of methods for integrative dimension reduction. Information Sciences. ISSN 0020-0255. 493, p. 105–119. doi: 10.1016/j.ins.2019.04.041. Full text in Research Archive Show summary

  Dimension reduction (DR) methods play an inevitable role in analyzing and visualizing high-dimensional multi-source data. In the recent decades many variants of these methods have been developed in various disciplines and domains. Due to the diversity and an ever-increasing number of developed techniques, choosing the right method for the given problem is a difficult task. In this study we benchmark 87 methods for integrative dimension reduction of mRNA expression and DNA methylation data, which is a common problem in biology and medicine. Our ranking is obtained based on four main factors: quality of dimension reduction (local, global, and local-global neighborhood preservation), clustering quality, speed and sensitivity to input parameters on multiple datasets generated by InterSIM (a semi-realistic multi-source data simulator in the genomics domain). The results are later validated on a real dataset for breast cancer through visual evaluation metrics such as co-ranking matrices, inspection of true cancer sub-types in two-dimensional projections, and LCMC curves. We also demonstrate the relationship between the methods via network inference. The findings in this study can be useful in algorithm selection and planning of experimental design in multi-source data analysis.

• Thoresen, Magne (2018). Er effekten forskjellig blant kvinner og menn? Tidsskrift for Den norske legeforening (Tidsskriftet). ISSN 0029-2001. 138, p. 1547–1547. doi: 10.4045/tidsskr.18.0602. Full text in Research Archive
• Thoresen, Magne (2018). Konfundering – et tilbakevendende problem. Tidsskrift for Den norske legeforening (Tidsskriftet). ISSN 0029-2001. 138, p. 257–257. doi: 10.4045/tidsskr.17.0795. Full text in Research Archive
• Buonaccorsi, John P.; Romeo, Giovanni & Thoresen, Magne (2018). Model-based bootstrapping when correcting for measurement error with application to logistic regression. Biometrics. ISSN 0006-341X. 74(1), p. 135–144. doi: 10.1111/biom.12730. Full text in Research Archive Show summary

  When fitting regression models, measurement error in any of the predictors typically leads to biased coefficients and incorrect inferences. A plethora of methods have been proposed to correct for this. Obtaining standard errors and confidence intervals using the corrected estimators can be challenging and, in addition, there is concern about remaining bias in the corrected estimators. The bootstrap, which is one option to address these problems, has received limited attention in this context. It has usually been employed by simply resampling observations, which, while suitable in some situations, is not always formally justified. In addition, the simple bootstrap does not allow for estimating bias in non‐linear models, including logistic regression. Model‐based bootstrapping, which can potentially estimate bias in addition to being robust to the original sampling or whether the measurement error variance is constant or not, has received limited attention. However, it faces challenges that are not present in handling regression models with no measurement error. This article develops new methods for model‐based bootstrapping when correcting for measurement error in logistic regression with replicate measures. The methodology is illustrated using two examples, and a series of simulations are carried out to assess and compare the simple and model‐based bootstrap methods, as well as other standard methods. While not always perfect, the model‐based approaches offer some distinct improvements over the other methods.

• Sørensen, Øystein; Hellton, Kristoffer Herland; Rattalma, Arnoldo Frigessi Di & Thoresen, Magne (2018). Covariate Selection in High-Dimensional Generalized Linear Models With Measurement Error. Journal of Computational And Graphical Statistics (JCGS). ISSN 1061-8600. 27(4), p. 739–749. doi: 10.1080/10618600.2018.1425626. Full text in Research Archive Show summary

  In many problems involving generalized linear models, the covariates are subject to measurement error. When the number of covariates p exceeds the sample size n, regularized methods like the lasso or Dantzig selector are required. Several recent papers have studied methods which correct for measurement error in the lasso or Dantzig selector for linear models in the p > n setting. We study a correction for generalized linear models, based on Rosenbaum and Tsybakov’s matrix uncertainty selector. By not requiring an estimate of the measurement error covariance matrix, this generalized matrix uncertainty selector has a great practical advantage in problems involving high-dimensional data. We further derive an alternative method based on the lasso, and develop efficient algorithms for both methods. In our simulation studies of logistic and Poisson regression with measurement error, the proposed methods outperform the standard lasso and Dantzig selector with respect to covariate selection, by reducing the number of false positives considerably. We also consider classification of patients on the basis of gene expression data with noisy measurements. Supplementary materials for this article are available online.

• Ghosh, Abhik & Thoresen, Magne (2017). Non-concave penalization in linear mixed-effect models and regularized selection of fixed effects. AStA Advances in Statistical Analysis. ISSN 1863-8171. p. 1–32. doi: 10.1007/s10182-017-0298-z. Full text in Research Archive Show summary

  Mixed-effect models are very popular for analyzing data with a hierarchical structure. In medical applications, typical examples include repeated observations within subjects in a longitudinal design, patients nested within centers in a multicenter design. However, recently, due to the medical advances, the number of fixed-effect covariates collected from each patient can be quite large, e.g., data on gene expressions of each patient, and all of these variables are not necessarily important for the outcome. So, it is very important to choose the relevant covariates correctly for obtaining the optimal inference for the overall study. On the other hand, the relevant random effects will often be low-dimensional and pre-specified. In this paper, we consider regularized selection of important fixed-effect variables in linear mixed-effect models along with maximum penalized likelihood estimation of both fixed and random-effect parameters based on general non-concave penalties. Asymptotic and variable selection consistency with oracle properties are proved for low-dimensional cases as well as for high dimensionality of non-polynomial order of sample size (number of parameters is much larger than sample size). We also provide a suitable computationally efficient algorithm for implementation. Additionally, all the theoretical results are proved for a general non-convex optimization problem that applies to several important situations well beyond the mixed model setup (like finite mixture of regressions) illustrating the huge range of applicability of our proposal. © 2017 Springer Verlag

• Hellton, Kristoffer Herland & Thoresen, Magne (2017). When and Why are Principal Component Scores a Good Tool for Visualizing High-dimensional Data? Scandinavian Journal of Statistics. ISSN 0303-6898. 44, p. 581–597. doi: 10.1111/sjos.12264. Full text in Research Archive Show summary

  Principal component analysis is a popular dimension reduction technique often used to visualize high‐dimensional data structures. In genomics, this can involve millions of variables, but only tens to hundreds of observations. Theoretically, such extreme high dimensionality will cause biased or inconsistent eigenvector estimates, but in practice, the principal component scores are used for visualization with great success. In this paper, we explore when and why the classical principal component scores can be used to visualize structures in high‐dimensional data, even when there are few observations compared with the number of variables. Our argument is twofold: First, we argue that eigenvectors related to pervasive signals will have eigenvalues scaling linearly with the number of variables. Second, we prove that for linearly increasing eigenvalues, the sample component scores will be scaled and rotated versions of the population scores, asymptotically. Thus, the visual information of the sample scores will be unchanged, even though the sample eigenvectors are biased. In the case of pervasive signals, the principal component scores can be used to visualize the population structures, even in extreme high‐dimensional situations. The final version of this research has been published in Scandinavian Journal of Statistics. © 2017 Wiley

• Thoresen, Magne (2017). Logistisk regresjon - anvendt og anvendelig. Tidsskrift for Den norske legeforening (Tidsskriftet). ISSN 0029-2001. 137(19). doi: 10.4045/tidsskr.17.0309. Full text in Research Archive
• Leder, Lena; Kolehmainen, Marjukka; Narverud, Ingunn; Dahlman, Ingrid; Myhrstad, Mari & Mello, Vanessa D. De [Show all 27 contributors for this article] (2016). Effects of a healthy Nordic diet on gene expression changes in peripheral blood mononuclear cells in response to an oral glucose tolerance test in subjects with metabolic syndrome: A SYSDIET sub-study. Genes & Nutrition. ISSN 1555-8932. 11. doi: 10.1186/s12263-016-0521-4. Full text in Research Archive Show summary

  Background: Diet has a great impact on the risk of developing features of metabolic syndrome (MetS), type 2 diabetes mellitus (T2DM), and cardiovascular diseases (CVD). We evaluated whether a long-term healthy Nordic diet (ND) can modify the expression of inflammation and lipid metabolism-related genes in peripheral blood mononuclear cells (PBMCs) during a 2-h oral glucose tolerance test (OGTT) in individuals with MetS. Methods: A Nordic multicenter randomized dietary study included subjects (n = 213) with MetS, randomized to a ND group or a control diet (CD) group applying an isocaloric study protocol. In this sub-study, we included subjects (n = 89) from three Nordic centers: Kuopio (n =26), Lund (n = 30), and Oulu (n = 33) with a maximum weight change of ±4 kg, high-sensitivity C-reactive protein concentration ≤10 mg L−1, and baseline body mass index <39 kg m−2. PBMCs were isolated, and the mRNA gene expression analysis was measured by quantitative real-time polymerase chain reaction (qPCR). We analyzed the mRNA expression changes of 44 genes before and after a 2hOGTT at the beginning and the end of the intervention. Results: The healthy ND significantly down-regulated the expression of toll-like receptor 4 (TLR4), interleukin 18 (IL18), and thrombospondin receptor (CD36) mRNA transcripts and significantly up-regulated the expression of peroxisome proliferator-activated receptor delta (PPARD) mRNA transcript after the 2hOGTT compared to the CD. Conclusions: A healthy ND is able to modify the gene expression in PBMCs after a 2hOGTT. However, more studies are needed to clarify the biological and clinical relevance of these findings.

• Buonaccorsi, John P.; Prochenka, Agnieszka; Thoresen, Magne & Ploski, Rafal (2016). Correcting for binomial measurement error in predictors in regression with application to analysis of DNA methylation rates by bisulfite sequencing. Statistics in Medicine. ISSN 0277-6715. 35(22), p. 3987–4007. doi: 10.1002/sim.6988. Full text in Research Archive
• Hellton, Kristoffer Herland & Thoresen, Magne (2016). Integrative clustering of high-dimensional data with joint and individual clusters. Biostatistics. ISSN 1465-4644. 17(3), p. 537–548. doi: 10.1093/biostatistics/kxw005. Full text in Research Archive Show summary

  When measuring a range of genomic, epigenomic, and transcriptomic variables for the same tissue sample, an integrative approach to analysis can strengthen inference and lead to new insights. This is also the case when clustering patient samples, and several integrative cluster procedures have been proposed. Common for these methodologies is the restriction to a joint cluster structure, equal in all data layers. We instead present a clustering extension of the Joint and Individual Variance Explained algorithm (JIVE), Joint and Individual Clustering (JIC), enabling the construction of both joint and data type-specific clusters simultaneously. The procedure builds on the connection between k-means clustering and principal component analysis, and hence, the number of clusters can be determined by the number of relevant principal components. The proposed procedure is compared with iCluster, a method restricted to only joint clusters, and simulations show that JIC is clearly advantageous when both individual and joint clusters are present. The procedure is illustrated using gene expression and miRNA levels measured in breast cancer tissue from The Cancer Genome Atlas. The analysis suggests a division into three joint clusters common for both data types and two expression-specific clusters. The final version of this research has been published in Biostatistics. © 2016 Oxford University Press

View all works in NVA

• Isaksen, Irja Minde; Skeie, Guri; Brustad, Magritt; Kværner, Ane Sørlie; Thoresen, Magne & Hjartåker, Anette (2026). Dietary protein source and colorectal cancer risk in a prospective cohort of Norwegian women: preliminary results. Full text in Research Archive
• Angelsen, Madeleine Elisabeth; Najar, Hemen; Svendsen, Mette; Thoresen, Magne; Peleikis, Dawn Elisabeth & Retterstøl, Kjetil (2026). Clinical dietitian-led nutrition counseling and exercise to reduce cardiovascular risk in adults living with a BMI above 27 and severe mental illness: the NORMI-Heart trial protocol. doi: 10.3389/fnut.2026.1700251. Full text in Research Archive
• Konglevoll, Dina Moxness; Andersen, Lene Frost; Thoresen, Magne; Totland, Torunn Holm; Hopstock, Laila Arnesdatter & Hjartåker, Anette [Show all 7 contributors for this article] (2025). Dietary trajectories over 21 years and frailty in Norwegian older adults. doi: 10.1007/s00394-024-03482-z. Full text in Research Archive Show summary

  This study investigated the association between dietary trajectories over two decades and frailty in older adults in the large Norwegian population-based study, the Tromsø Study. We found that there was an association between maintaining a moderately healthy to very healthy diet through adulthood and lower frailty in older age.

• Strømland, Stine Sofie; Aspholm, Trude; Fjørtoft, Jenny; Brønner, Charlotte Utheim; Stolt, Emma & Snipen, Lars-Gustav [Show all 12 contributors for this article] (2024). Effects of sulfur amino acid restriction on gut microbiota and inflammation in people with overweight and obesity: an 8-week randomized controlled dietary intervention study. Full text in Research Archive
• Konglevoll, Dina Moxness; Hjartåker, Anette; Hopstock, Laila Arnesdatter; Strand, Bjørn Heine; Thoresen, Magne & Andersen, Lene Frost [Show all 7 contributors for this article] (2023). Proteininntak og skrøpelighet hos eldre deltagere i Tromsøundersøkelsen (1994–2016). Full text in Research Archive
• Konglevoll, Dina Moxness; Andersen, Lene Frost; Hopstock, Laila Arnesdatter; Strand, Bjørn Heine; Thoresen, Magne & Totland, Torunn Holm [Show all 8 contributors for this article] (2023). Fish intake and pre-frailty in Norwegian older adults. A prospective cohort: the Tromsø Study 1994-2016. Full text in Research Archive
• Konglevoll, Dina Moxness; Andersen, Lene Frost; Hopstock, Laila Arnesdatter; Strand, Bjørn Heine; Thoresen, Magne & Totland, Torunn Holm [Show all 8 contributors for this article] (2023). Fish intake and pre-frailty in Norwegian older adults. A prospective cohort: the Tromsø Study 1994-2016. Full text in Research Archive
• Konglevoll, Dina Moxness; Andersen, Lene Frost; Hopstock, Laila Arnesdatter; Strand, Bjørn Heine; Thoresen, Magne & Totland, Torunn Holm [Show all 8 contributors for this article] (2023). Fish intake and pre-frailty in Norwegian older adults. A prospective cohort study: the Tromsø Study 1994-2016. Full text in Research Archive
• Konglevoll, Dina Moxness; Andersen, Lene Frost; Hopstock, Laila Arnesdatter; Strand, Bjørn Heine; Thoresen, Magne & Totland, Torunn Holm [Show all 8 contributors for this article] (2023). Fiskeinntak of pre-skrøpelighet hos eldre deltagere i Tromsøundersøkelsen: et prospektivt studie (1994-2016). Full text in Research Archive
• Konglevoll, Dina Moxness; Hjartåker, Anette; Hopstock, Laila Arnesdatter; Strand, Bjørn Heine; Thoresen, Magne & Andersen, Lene Frost [Show all 7 contributors for this article] (2022). Protein intake and the risk of pre-frailty/frailty in Norwegian older adults. The Tromsø Study 1994-2016. Full text in Research Archive
• Konglevoll, Dina Moxness; Hjartåker, Anette; Hopstock, Laila Arnesdatter; Strand, Bjørn Heine; Thoresen, Magne & Andersen, Lene Frost [Show all 7 contributors for this article] (2022). Protein intake and the risk of pre-frailty/frailty in Norwegian older adults. The Tromsø Study 1994-2016. Full text in Research Archive
• Konglevoll, Dina Moxness; Hjartåker, Anette; Hopstock, Laila Arnesdatter; Strand, Bjørn Heine; Thoresen, Magne & Andersen, Lene Frost [Show all 7 contributors for this article] (2022). Protein intake and the risk of pre-frailty/frailty in Norwegian older adults. The Tromsø Study 1994-2016. Full text in Research Archive
• Konglevoll, Dina Moxness; Hjartåker, Anette; Hopstock, Laila Arnesdatter; Strand, Bjørn Heine; Thoresen, Magne & Andersen, Lene Frost [Show all 7 contributors for this article] (2022). Protein intake and the risk of pre-frailty/frailty in Norwegian older adults. The Tromsø Study 1994-2016. Full text in Research Archive
• Christensen, Jacob Juel; Ulven, Stine Marie; Thoresen, Magne; Westerman, Kenneth; Holven, Kirsten Bjørklund & Andersen, Lene Frost (2020). Associations between dietary patterns and gene expression pattern in PBMCs: a cross-sectional study. Full text in Research Archive
• Rosseland, Leiv Arne; Reme, Silje Endresen; Simonsen, Tone Breines; Thoresen, Magne; Nielsen, Christopher Sivert & Eberhard-Gran, Malin (2020). Corrigendum to: Are labor pain and birth experience associated with persistent pain and postpartum depression? A prospective cohort study. Scandinavian Journal of Pain. ISSN 1877-8860. 20(4), p. 865–865. doi: 10.1515/sjpain-2020-2023. Full text in Research Archive
• Rosseland, Leiv Arne; Reme, Silje Endresen; Simonsen, Tone Breines; Thoresen, Magne; Nielsen, Christopher Sivert & Eberhard-Gran, Malin Marie Cecilia (2020). Reply: Response to letter to the editor "Labor pain, birth experience and postpartum depression". Scandinavian Journal of Pain. ISSN 1877-8860. 20(4), p. 861–863. doi: 10.1515/sjpain-2020-0130. Full text in Research Archive
• Rundblad, Amanda; Larsen, Sunniva Veen; Myhrstad, Mari; Ottestad, Inger; Thoresen, Magne & Holven, Kirsten Bjørklund [Show all 7 contributors for this article] (2019). Characterization of PBMC gene expression and lipoprotein subclasses among plasma TG responders and non-responders. Full text in Research Archive
• Thoresen, Magne (2019). Det problematiske oddsforholdet. Tidsskrift for Den norske legeforening (Tidsskriftet). ISSN 0029-2001. 139(7). doi: 10.4045/tidsskr.19.0011. Full text in Research Archive
• Larsen, Sunniva Veen; Holven, Kirsten Bjørklund; Flatberg, Arnar; Christensen, Jacob J.; Leder, Lena & Blomhoff, Rune [Show all 11 contributors for this article] (2018). Effects of replacing saturated fatty acids with polyunsaturated fatty acids on gene expression profiles of human peripheral blood mononuclear cells: a double-blind randomised controlled trial. Full text in Research Archive
• Hansson, Patrik; Holven, Kirsten Bjørklund; Øyri, Linn Kristin Lie; Brekke, Hilde Kristin; Biong, Anne Sofie & Gjevestad, Gyrd Omholt [Show all 8 contributors for this article] (2018). Postprandial effects of different dairy products on triglycerides in lean and overweight subjects: a randomized controlled cross-over study. Full text in Research Archive
• Hansson, Patrik; Holven, Kirsten Bjørklund; Øyri, Linn Kristin Lie; Brekke, Hilde Kristin; Biong, Anne Sofie & Gjevestad, Gyrd Omholt [Show all 8 contributors for this article] (2018). Postprandial effects of different dairy products on triglycerides in lean and overweight subjects: a randomized controlled cross-over study. Full text in Research Archive
• Hansson, Patrik; Holven, Kirsten Bjørklund; Øyri, Linn Kristin Lie; Brekke, Hilde Kristin; Gjevestad, Gyrd Omholt & Thoresen, Magne [Show all 7 contributors for this article] (2018). SEX DIFFERENCES IN POSTPRANDAL RESPONSE TO DIFFERENT DAIRY PRODUCTS ON LIPOPROTEIN SUBCLASSES: A RANDOMIZED CONTROLLED CROSS-OVER TRIAL. Full text in Research Archive
• Rundblad, Amanda; Larsen, Sunniva Veen; Myhrstad, Mari; Ottestad, Inger; Thoresen, Magne & Holven, Kirsten Bjørklund [Show all 7 contributors for this article] (2018). Gene expression changes in plasma triglyceride responders and non-responders to omega-3 supplementation. Full text in Research Archive
• Hansson, Patrik; Holven, Kirsten Bjørklund; Øyri, Linn Kristin Lie; Brekke, Hilde Kristin; Biong, Anne Sofie & Gjevestad, Gyrd Omholt [Show all 8 contributors for this article] (2017). Postprandial effects of different dairy products on blood lipids in lean and overweight subjects. Full text in Research Archive
• Bratsberg, Andrea Mathisen & Thoresen, Magne (2025). On variable selection in high-dimensional omics data in the presence of latent variables. Universitetet i Oslo. ISSN 9788234806582. Full text in Research Archive Show summary

  In biomedical sciences, advances in high-throughput technology have made it possible to measure large amounts of molecular data, such as gene expression levels or protein abundances, for every cell or organism in a study. This data, often referred to as "omics" data, holds the potential to uncover complex biological mechanisms underlying various diseases. However, a common challenge is that the number of molecular variables far exceeds the number of independent samples, making it difficult for traditional statistical methods to detect reliable patterns in the data. To analyze such high-dimensional data, it is generally assumed that either (1) only a small subset of variables is truly relevant (sparsity principle), or (2) the data can be represented in a lower-dimensional space. This thesis focuses on variable selection in high-dimensional datasets. The first part is concerned with modeling of longitudinal data, where each participant is observed at multiple time points. Here, it is important to account for the correlation between repeated observations from the same individual to perform effective variable selection. We propose a method based on mixed models to identify important variables while filtering out irrelevant ones. We prove that, under assumption (1), our method can successfully identify the relevant variables with high probability. The second part of the thesis also addresses correlation, but this time among the high-dimensional variables rather than between observations from the same individual. Here, we assume that both the predictors and the response variable are influenced by a shared set of unobserved latent variables. The number of latent variables is assumed to be much smaller than the number of predictors, thereby satisfying assumption (2). We explore variable selection in this context as well, and show that even without assuming sparsity, methods that produce sparse solutions can still provide good predictions.

• Hellton, Kristoffer Herland & Thoresen, Magne (2015). On high-dimensional principal component analysis in genomics: consistency and robustness. Universitetet i Oslo. ISSN 9788282649513. Full text in Research Archive Show summary

  Statistiker og sivilingeniør (MSc) Kristoffer H. Hellton har utvidet forståelsen av prinsipalkomponent-analyse (PCA), en sentral metode for å analysere høy-dimensjonale genetiske data. Arbeidet kan blant annet bidra til å forklare den paradoksale situasjonen hvor metoden har dårlige teoretiske egenskaper, men likevel fungerer svært godt i praksis. I tillegg presenteres det i avhandlingen en ny metode som integrerer flere ulike genetisk datatyper for å identifisere subgrupper av pasienter, slik at sykdomsbehandling i større grad kan tilpasses enkeltindividet. PCA brukes til å redusere dimensjonen av store datamengder. Dette gjøres ved å konstruere et sett med et fåtall såkalte scorer pr. observasjon, hvor disse beholder mest mulig av variasjonen i de originale dataene. Derfor er teknikken sentral i analyser av genetisk data, som f. eks. gen-uttrykk, der flere titusener av variabler måles samtidig. Hellton har vist at selv om metoden ikke kan forventes å gi helt korrekte estimater av scorene når antallet variabler er større enn antallet observasjoner (som er vanlig i genetiske data), vil visualiseringer basert på scorene kunne gi et svært godt bilde av den sanne strukturen i observasjonene. Dette gjør PCA godt egnet til å identifisere subgrupper av pasienter, også basert på flere høy-dimensjonale genetiske datatyper. I avhandlingen presenteres en utvidelse av metoden der både subgrupper felles for alle datatypene og subgrupper spesifikke for hver enkelt datatype kan identifiseres samtidig. Siden genetiske variabler er grunnleggende vanskelig å måle nøyaktig, har doktoranden også utforsket og kvantifisert effekten av teknisk målefeil på de ulike bestanddelene i PCA.

View all works in NVA