<p>Turniri su najvi&scaron;e istraživana klasa orijentisanih grafova. U tezi su prezentovana dva tipa rezultata. Prvi se odnosi na tzv. neizbežne podgrafove. Obuhvata Hamiltonove bajpase, podgrafove C(<em>n, i</em>) i alternativne Hamiltonove konture. Drugi se bavi problemima frekvencija skorova u običnim, bipartitnim i 3-partitnim turnirima.</p> / <p>Tournaments are the most investigated class of oriented graphs. Two type of results are presented in the thesis. First one is related to so called unavoidable subgraphs. It discusses Hamiltonian bypasses, subgraphs C(n, i) and antidirected Hamiltonian cycles. The second deals with problems of score frequencies in ordinary, bipartite and 3-partite tournaments.</p>

SAMMANFATTNING Titel: Sambandet mellan ESG-betyg och finansiell prestation inom energibranschen – en kvantitativ studie inom EU-länderna  Nivå: Examensarbete på grundnivå (kandidatexamen) i ämnet företagsekonomi  Författare: Fanny Bergström och Julia Larsson  Handledare: Jan Svanberg  Datum: 2023 - maj  Syfte: Intresset för hållbarhet och hållbarhetsredovisning har växt sig allt större och i takt med detta har sambandet mellan ett företags ESG-betyg och dess finansiella prestation blivit ett omdiskuterat forskningsområde. Tidigare forskning har kommit fram till motstridiga resultat då resultaten skiljer sig branscher och länder emellan. Med anledning av detta ska denna studie rikta in sig på energibranschen i EU för att utveckla det aktuella forskningsområdet. Studiens syfte är därmed att undersöka om det finns ett samband mellan företagens ESG-betyg och finansiell prestation inom energibranschen i EU. Studien ämnar dessutom undersöka vad ett eventuellt samband visar samt om sambandet skiljer sig om ett företag fokuserar mer individuellt på något av delkomponenterna miljö, social och bolagsstyrning.  Metod: Studien utgår ifrån en kvantitativ metod och tillämpar en deduktiv ansats. Datan som inhämtats är av sekundär art och har hämtats ifrån databasen Refinitiv Eikon. Urvalet består av 86 energibolag från 15 olika länder inom EU och har analyserats under en tidsperiod på fem år. Datan har analyserats med hjälp av univariat-, bivariat- och multivariat analys i statistikprogrammet SPSS.  Resultat och slutsats: Resultatet indikerade på ett negativt signifikant samband mellan ESG-betyg och ROA men också mellan E-betyget samt S-betyget och ROA. Fortsatt visade resultatet inget signifikant samband mellan G-betyget och ROA samtidigt som det inte, med varken ESG-betyget eller de enskilda komponenterna, fanns något signifikant samband med ROE.  Examensarbetets bidrag: Denna studie bidrar med forskning om sambandet mellan ESG-betyg och finansiell prestation i en bransch som är relativt outforskad. Den utforskar även sambandet mellan de tre enskilda komponenterna och finansiell prestation. På så sätt går det att få en ökad kunskap om sambandet även i denna bransch.  Förslag till fortsatt forskning: Förslag till vidare forskning har föreslagits utefter denna studies begränsningar. Ett förslag är att använda data från fler databaser än en för att då kunna öka studiens generaliserbarhet. Ytterligare ett förslag är att undersöka detta samband i fler länder och fler branscher för att på så sätt kunna få en större bild på hur sambandet ser ut i olika branscher världen över.  Nyckelord: ESG-betyg, E-, S- och G-betyg, ROA, ROE, Finansiell prestation och Energibranschen / ABSTRACT Title: The relationship between ESG-score and financial performance within the energy sector – a quantitative study within the EU countries.  Level: Student thesis, final assignment for Bachelor Degree in Business Administration  Author: Fanny Bergström and Julia Larsson  Supervisor: Jan Svanberg  Date: 2023 – May  Aim: The interest in sustainability and sustainability reporting has increased, and because of this, the relationship between a company´s ESG-rating and its financial performance has become a debated area of research. The result of previous research is conflicting as the results differ between sectors and countries. Due to this, this study will focus on the energy sector in the EU in order to develop the current research area. The purpose of the study is therefore to investigate whether there is a relationship between the companies ESG-rating and financial performance within the energy sector in the EU. The study also intends to investigate what a possible relationship shows and whether the relationship differs if a company focuses more individually on one of the sub-components environments, social and governance.  Method: The study is based on a quantitative method and has a deductive approach. The collected data is secondary and has been obtained from the Refinitiv Eikon database. The sample consists of 86 energy companies from 15 different countries in EU and has been analyzed over a period of five years. The data has been analyzed using univariate-, bivariate- and multivariate analysis in the statistical program SPSS.  Results and conclusions: The result indicates a negative significant correlation between ESG-score and ROA but also between E-score, S-score and ROA. The result showed a non-significant correlation between G-score and ROA while there was a non-significant correlation with ROE for either ESG-score or the individual components.  Contribution of the thesis: This study contributes to the research on the relationship between ESG-score and financial performance in a sector that is relatively unexplored. This study also explores the relationship between the three individual components and financial performance, thus allowing for a better understanding of the relationship in this sector.  Suggestions for future research: Suggestions for the future research have been proposed on the limitations of this study. One suggestion is to use data from more than one database to increase the generalizability of the study. Another suggestion is to investigate the relationship in more countries and sectors to get more knowledge about the relationship in different sectors and countries in the world.  Key Words: ESG-score, E-, S- and G-score, ROA, ROE, financial performance, energy sector

Estimating large covariance matrices from small samples is an important problem in many fields. Among others, this includes spatial statistics and data assimilation. In this thesis, we deal with several methods of covariance estimation with emphasis on regula- rization and covariance models useful in filtering problems. We prove several properties of estimators and propose a new filtering method. After a brief summary of basic esti- mating methods used in data assimilation, the attention is shifted to covariance models. We show a distinct type of hierarchy in nested models applied to the spectral diagonal covariance matrix: explicit estimators of parameters are computed by the maximum like- lihood method and asymptotic variance of these estimators is shown to decrease when the maximization is restricted to a subspace that contains the true parameter value. A similar result is obtained for general M-estimators. For more complex covariance mo- dels, maximum likelihood method cannot provide explicit parameter estimates. In the case of a linear model for a precision matrix, however, consistent estimator in a closed form can be computed by the score matching method. Modelling of the precision ma- trix is particularly beneficial in Gaussian Markov random fields (GMRF), which possess a sparse precision matrix. The...

No description available.

Le diagnostic de l'asthme professionnel (AP) est toujours un défi. Le test de provocation bronchique spécifique (TPS), comme une méthode de diagnostic de référence, n’est pas aisément accessible. Cette étude diagnostique rétrospective vise à évaluer des outils diagnostiques actuels et à développer des scores cliniques pour AP (définis comme ayant le résultat positif en TPS). Les données concernant les travailleurs soupçonnés d'avoir de l’AP qui, d’une part, ont été exposés aux agents de haut-poids-moléculaire élevé (HPM) (n=139) et à bas-poids-moléculaire (BPM) (n=285), et d’autre part, ont travaillé encore un mois avant de l’évaluation de TPS. Par ailleurs, les modèles de régression logistique sont développés dans chaque groupe d'exposition. Ainsi, concernant des tests objectifs, les valeurs de différents tests distinctifs sont ajoutées aux caractéristiques cliniques, et enfin, le résultat a été évalué. Les modèles ont été testés pour l’exactitude, et pour la validation interne par la procédure bootstrapping. Suite à cela, les modèles finaux sont traduits en scores cliniques et le score total est stratifié en groupes à risque. Chez les travailleurs exposés à des agents BPM, si le test de la méthacholine est fait isolément, le modèle prédictif n'a pas montré de meilleures valeurs diagnostiques que le test de provocation. Cependant, dans le groupe HPM, le modèle final, y compris le sexe, l'âge> 40 ans, la durée des symptômes ≥1 an, la rhinoconjonctivite, l'utilisation de corticostéroïdes inhalés, le test de provocation à la méthacholine, et le test de la piqûre épidermique spécifique, avait un bon calibrage et une validation interne raisonnable. Par ailleurs, la catégorie de sujets avec une probabilité élevée d’avoir AP avait une meilleure spécificité et une meilleure valeur prédite positive par rapport à la combinaison de test de provocation à la méthacholine et de la piqûre épidermique spécifique dans la détection de l'AP, cependant n'avait pas de signification statistique. En conclusion, ce modèle quantifie la probabilité individuelle d'AP. Dans les centres où l'accès à TPS est difficile ou impossible, notre modèle serait utile dans le diagnostic d’OA, néanmoins, la validation externe du modèle reste nécessaire. Mots-clés : asthme professionnel, modèle diagnostique, prévention, score clinique. / The diagnosis of occupational asthma (OA) is challenging since the use of specific inhalation challenge (SIC) as the reference test is not widely accessible. This retrospective diagnostic study is aimed to evaluate current diagnostic tools and to develop clinical scores for OA (defined as positive SIC). Data from workers with suspected OA who were exposed to high-molecular-weight (HMW) (n=139) and low-molecular-weight (LMW) agents (n=285) and still working one month before the SIC were evaluated. Logistic regression models were developed in each exposure group. The added values of different objective tests to clinical and exposure characteristics were evaluated. The models were tested for accuracy, and, validated internally by the bootstrapping procedure. The final models were translated into clinical score and the sum scores were stratified into risk groups. In workers exposed to LMW agents, the predictive model did not perform better diagnostically than the methacholine challenge test alone. In the HMW group, the final model including sex, age >40 years, symptom duration ≥1 year, rhinoconjunctivitis, inhaled corticosteroid use, the methacholine challenge test, and specific SPT had a good accuracy and reasonable internal validation. The high probability category of the predictive model had a better specificity and positive predicted value compared to the combination of methacholine challenge test and specific SPT in detecting OA but did not reach the statistical significance. Our results suggest that this model could quantify an individual’s probability of OA. This model emphasizes the necessity of performing both tests in order to have a more accurate diagnosis in workers exposed to HMW agents. In centers where access to SIC is difficult or impossible, our model might be of benefit in diagnosing OA. Nevertheless, external validation of the model is necessary. Key words: occupational asthma, diagnostic model, prevention, clinical score.

Introduction : La fibrillation auriculaire (FA) est l’arythmie la plus répandue au monde avec une prévalence estimée autour de 1-2% pour la population générale. Il s’agit d’une maladie complexe de causes multifactorielles, telles que la génétique, l’environnement et les habitudes de vie. Il est aussi important de prendre en compte que les risques de FA augmentent drastiquement avec l’âge. La fibrillation auriculaire post-opératoire (POAF) est la complication la plus commune à la suite d’une chirurgie cardiaque. Elle est associée avec une prolongation de la durée d’hospitalisation et à une augmentation du risque d’accident vasculaire cérébral (AVC) et de la mortalité. Il devient donc une priorité d’identifier les individus à risque de POAF et d’AVC causés par la FA, surtout en considérant l’existence d’interventions simples qui peuvent atténuer ces risques, comme certaines thérapies pharmacologiques. Hypothèse : L’hypothèse de ce projet est qu’un score polygénique (PGS) aurait un impact sur l’identification des individus à plus haut risque de FA à la suite d’une chirurgie cardiaque, mais aussi sur l’identification des individus à risque de récidive de FA lors d’un premier épisode de FA après une chirurgie. Objectifs : Ce projet se divise en 2 objectifs principaux. Dans un premier temps, nous cherchons à évaluer la capacité prédictive d’un PGS déjà validé (mais jamais utilisé pour la POAF) pour la POAF après une chirurgie cardiaque dans une population québécoise. Dans un second temps, nous cherchons à évaluer la capacité prédictive du même PGS au niveau de la récidive à distance de la FA chez le sous-groupe de personnes qui ont eu un épisode de POAF. Méthodes : Nous avons inclus 2340 participants de la Biobanque hospitalière de l’Institut de Cardiologie de Montréal (ICM) qui ont subi une chirurgie cardiaque entre 2005 et 2020 et qui n’avaient pas de FA diagnostiquée avant la chirurgie. La POAF a été définie comme une FA survenant jusqu’à 30 jours après la chirurgie. Le génotypage de la cohorte a été effectué à l’aide d’une puce de génotypage pangénomique, suivi d’une imputation avec l’aide du panel de référence TOPMed. Le PGS déjà validé et corrigé pour les principales composantes génétiques (PGS-AF) a été calculé à l’aide des poids d’un PGS publié (identifiant du PGS catalog : PGS000016). L’association entre le PGS-AF et la POAF a été évaluée par une régression logistique avec et sans la correction pour les facteurs de risque cliniques connus de la FA, y compris le score de risque clinique CHARGE-AF. Pour valider la valeur prédictive du PGS-AF indépendamment des prédicteurs cliniques, un indice de reclassement net (NRI) a été calculé. L’association entre le PGS-AF et le CHARGE-AF au niveau de la récidive de FA à distance, à plus de 30 jours après la chirurgie, au sein du sous-groupe de patients atteints de POAF a été évaluée avec un modèle de régression de Cox. Résultats : Sur les 2340 participants inclus dans l’étude rétrospective (80% d’hommes; âgés de 66 ans [59-71] au moment de la chirurgie cardiaque), 871 (37%) ont développé de la POAF. La POAF était plus fréquente après une chirurgie valvulaire que les autres chirurgies cardiaques (43% contre 32%, P<0,001) et avec une augmentation dans le score de risque clinique CHARGE-AF (P<0,001). Le PGS-AF était significativement associé à la POAF (P<0,001; augmentation du risque de POAF de 46% par augmentation de 1 écart-type du PGS-AF; statistique C = 0,61). L’incidence de la POAF était significativement plus élevée chez les patients avec PGS-AF au-dessus du 95e percentile (62%) comparativement aux patients avec PGS-AF en dessous du 95% percentile (36%; ratio de cote 2.3, intervalle de confiance 95% 1.6-3.4; P<0,001). Dans un modèle combinant le PGS-AF avec des prédicteurs cliniques (CHARGE-AF et chirurgie valvulaire), l’association du PGS-AF avec la POAF reste significative (P<0,001). De plus, par rapport à la prédiction clinique uniquement, l’ajout du PGS-AF entraine une amélioration significative du modèle (statistique C de 0,68 contre 0,65; test du rapport de vraisemblance P<0,001; NRI = 35%). Parmi les patients atteints de POAF, 235 (27%) ont développé de la FA au cours d’un suivi médian de 4,4 ans. Le PGS-AF et le CHARGE-AF étaient à la fois associés de manière significative et indépendante à la récidive de FA à distance (P<0,05), où chaque augmentation d’écart-type du PGS-AF augmente le risque de récidive de FA de 19%. Discussion et conclusion : Un score de risque polygénique précédemment validé pour la fibrillation auriculaire (PGS-AF) est significativement associé à la FA survenant comme une complication de la chirurgie cardiaque, indépendamment des prédicteurs de risque cliniques. Bien que la capacité discriminative soit modeste globalement, le PGS-AF permet d’identifier 5% de la population de patients avec risque de POAF 2,3 fois plus élevé, une magnitude d’effet de pertinence clinique. Le PGS-AF est également associé à la récidive à distance de la FA et, s’il est validé, pourrait aider à identifier les patients plus susceptibles de bénéficier d’une anticoagulation à long terme pour prévenir les AVC. Plus globalement, ces données impliquent le risque polygénique pour expliquer la variance de scénarios cliniques complexes tels que les complications chirurgicales. / Introduction: Atrial fibrillation (AF) is the most common arrhythmia in the world with an estimated prevalence of around 1-2% of the general population. It is a complex disease with multifactorial causes, such as genetics, environment, and lifestyle. It is also important to consider that the risk of AF increases drastically with age. Postoperative atrial fibrillation (POAF) is the most common complication following cardiac surgery. It is associated with a prolonged hospital stay length and an increased risk of cerebrovascular accident (CVA) and mortality. It therefore becomes a priority to identify individuals at higher risk of POAF and stroke caused by AF, especially considering the existence of simple interventions that can mitigate these risks, such as certain pharmacological therapies. Hypothesis: The project hypothesis is that a polygenic score (PGS) would have an impact on the identification of individuals at higher risk of AF following cardiac surgery, but also on the identification of individuals at higher risk of recurrence of AF within the subgroup of patients who had a first episode of AF after surgery. Aims: This project is divided into 2 main objectives. First, we seek to assess the predictive ability of an already validated PGS (but never used before for POAF) for POAF after cardiac surgery in a Quebec population. In a second step, we seek to assess the predictive capacity of the same PGS for distant recurrence of AF within the subgroup of people who have had an episode of POAF. Methods: We included 2,340 participants from the Montreal Heart Institute (MHI) hospital Biobank who underwent heart surgery from 2005 to 2020 and did not have an AF diagnosis before surgery. POAF was defined as AF occurring in up to 30 days after surgery. Cohort genotyping was performed using a genome-wide genotyping array, followed by imputation using the TOPMed reference panel. The PGS already validated and corrected for the main genetic components (PGS-AF) was calculated using weights of a published PGS (PGS catalog ID: PGS000016). The association between PGS-AF and POAF was assessed by logistic regression with and without correction for known AF clinical risk factors, including the CHARGE-AF clinical score. To validate the predictive ability of PGS-AF independently of clinical predictors, a net reclassification index (NRI) was calculated. The association between PGS-AF and CHARGE-AF in remote AF recurrence, more than 30 days after cardiac surgery, within the subgroup of patients with POAF was assessed using a Cox Proportional Hazards Model. Results: Of the 2,340 participants included in the retrospective study (80% males; aged 66 years old [59-71] at the time of cardiac surgery), 871 (37%) developed POAF. POAF was more common after valve surgery than other cardiac surgeries (43% vs 32%, P<0.001) and with an increase in the CHARGE-AF clinical risk score (P<0.001). PGS-AF was significantly associated with POAF (P<0.001; POAF risk increased by 46% per each standard deviation increase in PGS-AF; C-statistic = 0.61). The incidence of POAF was significantly higher in patients with PGS-AF above the 95th percentile (62%) compared to patients with PGS-AF below the 95% percentile (36%; odds ratio 2.3, range 95% confidence interval 1.6-3.4; P<0.001). In a model combining PGS-AF with clinical predictors (CHARGE-AF and valve surgery), the association of PGS-AF with POAF remains significant (P<0.001). Moreover, compared to the clinical predictors only, the addition of PGS-AF leads to a significant improvement in the model (C-statistic of 0.68 vs 0.65; likelihood ratio test P<0.001; NRI = 35%). Among patients with POAF, 235 (27%) developed AF during a median follow-up of 4.4 years. PGS-AF and CHARGE-AF were both significantly and independently associated with distant recurrence of AF (P<0.05 for both), where each standard deviation increase in PGS-AF increases the risk of distant AF recurrence by 19%. Discussion and conclusion: A previously validated polygenic risk score for atrial fibrillation (PGS-AF) is significantly associated with AF occurring as a complication of cardiac surgery, independently of clinical risk predictors. Although the discriminative ability is modest overall, the PGS-AF identifies 5% of the patient population with a 2.3 times higher risk of POAF, an effect size of clinical relevance. PGS-AF is also associated with distant recurrence of AF and, if validated, could help identify patients who could most likely benefit from long-term anticoagulation to prevent stroke. More broadly, these data implicate polygenic risk to explain the variance of complex clinical scenarios such as surgical complications.

<p dir="ltr">Heat stress is one of the leading welfare concerns for modern swine production globally. Swine are especially susceptible as the species has inactive sudoriferous glands and, as a result, these animals thermoregulate mainly through behaviors such as wallowing and rooting up cooler patches of ground. In commercial settings, access to open ground or wallowing pits would present a threat to the animals’ health in the form of a vector for diseases and parasites. This puts aspects of swine welfare in direct conflict with one another, while public legislature and consumer pressures increasingly call for approaches to animal welfare that encompass all aspects of welfare to be considered. Efforts towards improved welfare are also crucial to the environmental and economical sustainability for producers and consumers. Previous genetic selection tactics used in swine herds focused intensely on a few traits and have led to modern animals being highly productive but also more environmentally sensitive. Intensely selected animals have energy margins that are narrower and more devoted to higher production leaving less accommodations for resilience in moments of stress. Since swine behaviorally thermoregulate, the use of behavior studies combined with those of genetics is one possible way to effectively identify heritable aspects of climatic resilience into future herds while still improving production traits. The primary objectives of this thesis were to develop an ethogram that can be used to base phenotypes of behavior; calculate the variance components of the behavior traits; correlate these components with those of maternal productivity and heat tolerance; and finally explore the genetic background of the behavior traits based on the identification of genomic markers significantly associated with them. Data was collected on 1,678 Landrace × Large White sows from a commercial swine herd in North Carolina. The data comprised information on climatic factors, litter performance, and behavior during a standardized human handling procedure in the summer of 2021. These animals were genotyped using the PorcineSNP50K Bead Chip (Illumina, San Diego, CA, USA). These studies describe the development of an ethogram for use on lactating sows in a commercial setting in order to create the novel behavior traits of Responsiveness Score (RS), Vocalization Score (VS), and Shave Time (ST). The genetic variance components were then estimated, and heritabilities found to be 0.17 ± 0.05 for RS, 0.15 ± 0.05 for VS, and 0.10 ± 0.05. These novel behavior traits had low genetic correlations with traits of maternal performance and heat stress resilience. Genome-wide association studies were then performed, and 19 genomic markers were significantly associated with the three behavior traits. However, only one region located on chromosome 3 harbors a candidate gene that may play a role in heat stress resilience. Although the novel behavior traits explored in this study are heritable, they were not found to be viable indicator traits for heat stress resilience due to their low genetic correlations with direct indicators of heat tolerance in lactating sows. It is important that additional phenotypic records of behaviors related to welfare are generated in the future to better map their polygenic nature. </p>

Background The number of liver resections is constantly rising over the last decades. Despite the reduction of overall mortality and morbidity in liver surgery, biliary leakage is still a relevant postoperative complication that can lead to a fatal postoperative course. Aim of this analysis is the identification of specific risk factors for postoperative biliary complications after liver resections and the development of a predictive biliary leakage risk score. Methods A single-center, retrospective analysis of 844 liver resections performed in the Department of Visceral, Thoracic and Vascular Surgery, Technische Universität Dresden, between 1/2013 and 12/2019 is conducted to identify risk factors for postoperative biliary leakage and a risk score for biliary leakage after hepatectomy is established based on multivariate regression. The score has been validated by an independent validation cohort consisting of 142 patients. Results Overall morbidity is 43.1% with 36% surgical complications and an overall mortality of 4.3%. Biliary leakage occurred in 15.8% of patients. A predictive score for postoperative biliary leakage based on age, major resection, pretreatment with FOLFOX/cetuximab and operating time is created. Patients are stratified to low (< 15%) and high (> 15%) risk with a sensitivity of 67.4% and a specificity of 70.7% in development cohort and a specificity of 68.2% and sensitivity of 75.8% in validation cohort. Conclusions The presented score is robust and has been validated in an independent patient cohort. Depending on the calculated risk, prevention or early treatment can be initiated to avoid bile leakage and to improve postoperative course.

This study aims to construct a state-of-the-art classifier model and compare it against a largelanguage model. A variation of SVM called LinearSVC was utilised and the BERT model usingbert-base-uncased was used. The data, provided by the Swedish Research Council, consisted ofresearch grant applications. The research grant applications were divided into two groups, whichwere further divided into several subgroups. The subgroups represented research fields such ascomputer science and applied physics. Significant class imbalances were present, with someclasses having only a tenth of the applications of the largest class. To address these imbalances,a new dataset was created using data that had been randomly oversampled. The models weretrained and tested on their ability to correctly assign a subgroup to a research grant application.Results indicate that the BERT model outperformed the SVM model on the original dataset,but not on the balanced dataset . Furthermore, the BERT model’s performance decreased whentransitioning from the original to the balanced dataset, due to overfitting or randomness. / Denna studie har som mål att bygga en state-of-the-art klassificerar model och sedan jämföraden mot en stor språkmodel. SVM modellen var en variation av SVM vid namn LinearSVC ochför BERT användes bert-base-uncased. Data erhölls från Vetenskapsrådet och bestod av forskn-ingsbidragsansökningar. Forskningsbidragsansökningarna var uppdelade i två grupper, som varytterligare uppdelade i ett flertal undergrupper. Dessa undergrupper representerar forsknings-fält såsom datavetenskap och tillämpad fysik. I den data som användes i studien fanns storaskillnader mellan klasserna, där somliga klasser hade en tiondel av ansökningarna som de storaklasserna hade. I syfte att lösa dessa klassbalanseringsproblem skapades en datamängd somundergått slumpmässig översampling. Modellerna tränades och testades på deras förmåga attkorrekt klassificera en forskningsbidragsansökan in i rätt undergrupp. Studiens fynd visade attBERT modellen presterade bättre än SVM modellen på både den ursprungliga datamängden,dock inte på den balanserade datamängden. Tilläggas kan, BERTs prestanda sjönk vid övergångfrån den ursprungliga datamängden till den balanserade datamängden, något som antingen berorpå överanpassning eller slump.

Yes / In English hospitals, the patient’s vital signs are monitored and summarised into a National Early Warning Score (NEWS). NEWS is more accurate than the quick sepsis related organ failure assessment (qSOFA) score at identifying patients with sepsis. We investigate the extent to which the accuracy of the NEWS is enhanced by developing computer-aided NEWS (cNEWS) models. We compared three cNEWS models (M0=NEWS alone; M1=M0 + age + sex; M2=M1 + subcomponents of NEWS + diastolic blood pressure) to predict the risk of sepsis. Methods: All adult emergency medical admissions discharged over 24-months from two acute hospitals (YH–York Hospital for model development; NH–Northern Lincolnshire and Goole Hospital for external model validation). We used a validated Canadian method for defining sepsis from administrative hospital data. Findings: The prevalence of sepsis was lower in YH (4.5%=1596/35807) than NH (8.5%=2983/35161). The c-statistic increased across models (YH: M0: 0.705, M1:0.763, M2:0.777; NH:M0: 0.708, M1:0.777, M2:0.791). At NEWS 5+, sensitivity increased (YH: 47.24% vs 50.56% vs 52.69%; NH: 37.91% vs 43.35% vs 48.07%)., the positive likelihood ratio increased (YH: 2.77 vs 2.99 vs 3.06; NH: 3.18 vs 3.32 vs 3.45) and the positive predictive value increased (YH: 11.44% vs 12.24% vs 12.49%; NH: 22.75% vs 23.55% vs 24.21%). Interpretation: From the three cNEWS models, Model M2 is the most accurate. Since it places no additional data collection burden on clinicians and can be automated, it may now be carefully introduced and evaluated in hospitals with sufficient informatics infrastructure. / The Health Foundation, National Institute for Health Research (NIHR) Yorkshire and Humberside Patient Safety Translational Research Centre / Research Development Fund Publication Prize Award winner, April 2019.