abstract: Several ways exist to improve pavement performance over time. One suggestion is to tailor the asphalt pavement mix design according to certain specified specifications, set up by each state agency. Another option suggests the addition of modifiers that are known to improve pavement performance, such as crumb rubber and fibers. Nowadays, improving asphalt pavement structures to meet specific climate conditions is a must. In addition, time and cost are two crucial settings and are very important to consider; these factors sometimes play a huge role in modifying the asphalt mix design needed to be set into place, and therefore alter the desired pavement performance over the expected life span of the structure. In recent studies, some methods refer to predicting pavement performance based on the asphalt mixtures volumetric properties. In this research, an effort was undertaken to gather and collect most recent asphalt mixtures’ design data and compare it to historical data such as those available in the Long-Term Pavement Performance (LTPP), maintained by the Federal Highway Administration (FHWA). The new asphalt mixture design data was collected from 25 states within the United States and separated according to the four suggested climatic regions. The previously designed asphalt mixture designs in the 1960’s present in the LTPP Database implemented for the test sections were compared with the recently designed pavement mixtures gathered, and pavement performance was assessed using predictive models. Three predictive models were studied in this research. The models were related to three major asphalt pavement distresses: Rutting, Fatigue Cracking and Thermal Cracking. Once the performance of the asphalt mixtures was assessed, four ranking criteria were developed to support the assessment of the mix designs quality at hand; namely, Low, Satisfactory, Good or Excellent. The evaluation results were reasonable and deemed acceptable. Out of the 48 asphalt mixtures design evaluated, the majority were between Satisfactory and Good. The evaluation methodology and criteria developed are helpful tools in determining the quality of asphalt mixtures produced by the different agencies. They provide a quick insight on the needed improvement/modification against the potential development of distress during the lifespan of the pavement structure. / Dissertation/Thesis / Masters Thesis Civil, Environmental and Sustainable Engineering 2020

The aim of the thesis is to research the methodological approaches used for reducing traffic accidents. One part of the methods is focused on methods of evaluating traffic accident rate on roads in terms of macro-analysis. Another group of methods for the prediction of traffic accidents. Subsequently, in the practical part the methodology is applied on two selected sections of roads and the resulting values are then compared with the actual dates that can explanatory power of these methods. Another part relates to the economic evaluation of road accidents on selected sections of roads and applications of integral indicators.

A reliable evaluation system is essential for every automatic process. While techniques for automatic segmentation of images have been extensively researched in recent years, evaluation of the same has not received an equal amount of attention. Amra Medical AB has developed a system for automatic segmentation of magnetic resonance (MR) images of human bodies using an atlas-based approach. Through their software, Amra is able to derive body composition measurements, such as muscle and fat volumes, from the segmented MR images. As of now, the automatic segmentations are quality controlled by clinical experts to ensure their correctness. This thesis investigates the possibilities to leverage predictive modelling to reduce the need for a manual quality control (QC) step in an otherwise automatic process. Two different regression approaches have been implemented as a part of this study: body composition measurement prediction (BCMP) and manual correction prediction (MCP). BCMP aims at predicting the derived body composition measurements and comparing the predictions to actual measurements. The theory is that large deviations between the predictions and the measurements signify an erroneously segmented sample. MCP instead tries to directly predict the amount of manual correction needed for each sample. Several regression models have been implemented and evaluated for the two approaches. Comparison of the regression models shows that local linear regression (LLR) is the most performant model for both BCMP and MCP. The results show that the inaccuracies in the BCMP-models, in practice, renders this approach useless. MCP proved to be a far more viable approach; using MCP together with LLR achieves a high true positive rate with a reasonably low false positive rate for several body composition measurements. These results suggest that the type of system developed in this thesis has the potential to reduce the need for manual inspections of the automatic segmentation masks.

Background: In today's digital world, there is increasing importance on cross-platform performance testing and the challenges faced by businesses in achieving efficient performance for applications across multiple platforms. Predictive models, such as machine learning and regression, have emerged as potential solutions to predict performance to be quickly analyzed, thus eliminating the need to execute an entire environment. Predicting performance can help firms save time and resources to keep pace with market demand, but potential risks and limitations need to be considered. With the increasing availability of data, predictive models have become effective problem-solving methods in various industries, including the testing industry. Objectives: This research aims to investigate the economic consequences and opportunities of implementing predictive models to predict cross-platform performance for firms operating in the software market and evaluate the performance of three models when predicting cross-platform performance. The study aims to add arguments to help businesses make informed decisions on the adoption of predictive models. Methods: The methodology employed in this research involved evaluating Multiple Linear Regression, Multiple Neural Network, and Random Forest, to gain insight into how such models perform when predicting performance. In addition to this analysis, interviews were conducted with industry experts to get an understanding of current processes and the potential benefits of adopting predictive models to identify the economic consequences of implementing such models. Results: The result shows that Multiple Linear Regression was the most promising one, with an R2 value of 0.79. Additionally, the research revealed that the current testing process faces difficulties when testing on multiple platforms. While predicting performance can provide cost and time savings, challenges and risks, such as data privacy and model trust, must also be considered. Conclusions: Multiple Linear Regression exhibited the most favorable performance among the evaluated models, with consistent results across all test runs and indicating a linear relationship. The economic consequences identified were the continuously required maintenance and updates of predictive models to remain accurate throughout the lifecycle. This implies ongoing costs, such as the complexity and cost of generating and storing the necessary data to train the models. Thus, the adoption of predictive models is still in its early stages, and while there are significant benefits, there are also challenges to address.

[pt] Empresas do ramo da Saúde vêm evoluindo seus modelos de gestão, desenvolvendo programas proativos para melhorar a qualidade e a eficiência dos seus serviços considerando informações históricas. Estratégias proativas buscam prevenir e detectar doenças precocemente e também melhorar os resultados das internações. Nesse sentido, uma tarefa desafiadora é identificar quais pacientes devem ser incluídos em programas proativos de saúde. Para isso, a previsão e a modelagem de variáveis relacionadas aos custos estão entre as abordagens mais amplamente utilizadas, uma vez que essas variáveis sào potenciais indicadores do risco, da gravidade e do consumo de recursos médicos de uma internação. A maioria das pesquisas nesta área têm como foco modelar variáveis de custo em uma perspectiva geral e prever variações de custos para períodos específicos. Por outro lado, este trabalho se concentra na previsão dos custos de um evento específico. Em particular, esta dissertação prescreve uma solução para a predição de internações de alto custo, visando dar apoio a gestores de serviços em saúde em suas ações proativas. Para esse fim, foi seguida a metodologia de pesquisa Design Science Research (DSR), aliada ao ciclo de vida de projeto de Ciência de Dados, sobre um cenário real de uma empresa de consultoria em saúde. Os dados fornecidos descrevem internações de pacientes através de suas características demográficas e do histórico de consumo de recursos médicos. Diferentes técnicas estatísticas e de Aprendizado de Máquina foram aplicadas, como Ridge Regression (RR), Least Absolute Shrinkage and Selection Operator (LASSO), Classification and Regression Trees (CART), Random Forest (RF) e Extreme Gradient Boosting (XGB). Os resultados experimentais evidenciaram que as técnicas RF e XGB apresentaram o melhor desempenho, atingindo AUCPR de 0,732 e 0,644, respectivamente. O modelo de predição da técnica RF foi capaz de detectar até 72 porcento, em média, das internações de alto custo com 33 porcento de precisão, o que representa 78,7 porcento do custo total gerado por tais internações. Além disso, os resultados monstraram que o uso de custo prévio e variáveis agregadas de consumo de recursos aumentaram a capacidade de predição do modelo / [en] Healthcare providers are evolving their management models, developing proactive programs to improve the quality and efficiency of their health services, considering the available historical information. Proactive strategies seek not only to prevent and detect diseases but also to enhance hospitalization outcomes. In this sense, one of the most challenging tasks is to identify which patients should be included in proactive health programs. To this end, forecasting and modeling cost-related variables are among the most widely used approaches for identifying such patients, since these variables are potential indicators of the patients hospitalization risk, their severity, and their medical resources consumption. Most of the existing research works in this area aim to model cost variables from an overall perspective and predict cost variations for specific periods. In contrast, this work focuses on predicting the costs of a particular event. Specifically, this thesis prescribes a solution for identifying high-cost hospitalizations, to support health service managers in their proactive actions. To this end, the Design Science Research (DSR) methodology was combined with the Data Science life cycle in a real scenario of a health consulting company. The data provided describes patients hospitalizations through their demographic characteristics and their medical resource consumption. Different statistical and Machine Learning techniques were used to predict high-cost hospitalizations, such as Ridge Regression (RR), Least Absolute Shrinkage and Selection Operator (LASSO), Classification and Regression Trees (CART), Random Forest (RF), and Extreme Gradient Boosting (XGB). The experimental results showed that RF and XGB presented the best performance, reaching an Area Under the Curve Precision-Recall (AUCPR) of 0.732 and 0.644, respectively. In the case of RF, the model was able to detect, on average, 72 percent of the high-cost hospitalizations with a 33 percent of Precision, which represents 78.7 percent of the total cost generated by the high-cost hospitalizations. Moreover, the obtained results showed that the use of prior cost and aggregated variables of resource consumption increased the model s ability to predict high-cost hospitalizations.

År 2019 installerades 732 solpaneler på taket i ett hus i Örebro universitet. Energiproduktionenav anläggningen samlades i en databas i Akademiska Hus med ett antal parametrar från enväderstation i samma hus. Att kunna modellera den här produktionen som en funktion avväderparametrar och historiska värden med hjälp av maskininlärning, och jämföra olikamodeller är målet i detta projekt. Det finns gjorda arbeten med samma mål i olikalaborationsmiljöer och andra platser men inte för denna anläggning. Mätvärden under två årfrån 2019 till 2021 kommer från Akademiska Hus och resultaten blir två modeller: ett NarrowNeural Network samt en Support Vector Machine med 7 procent avvikelse och en NonlinearAutoregressive Neural Network för envariatmodellen. / In 2019, 732 solar panels were installed on the roof of a building at Örebro University. Thesolar power production of the facility has been collected in a database in Akademiska Hus,along with several parameters from a weather station in the same building. The goal of thisproject is to model solar power production as a function of weather parameters and historicalvalues using machine learning techniques. This study investigates various predictive models tofind a suitable model for predicting this production. There have been several studies in theliterature that have performed this goal in various laboratory environments and other places,but not for this facility. The measured data for this study is recorded by Akademiska Hus forover two years from 2019 to 2021. The results of this work lead to two suitable machine learningmodels while using weather parameters: 1) Narrow Neural Network and 2) Support VectorMachine with 7% errors in both models. Moreover, this study has investigated univariatemodels to predict the solar power production as a time series based on its historical data. Forthis aim, a Nonlinear Autoregressive Neural Network has been applied which results inconsiderably low errors in the evaluations.

No / Drug vehicles are chemical carriers that aid a drug's passage through an organism. Whilst they possess no intrinsic efficacy they are designed to achieve desirable characteristics which can include improving a drug's permeability and or solubility, targeting a drug to a specific site or reducing a drug's toxicity. All of which are ideally achieved without compromising the efficacy of the drug. Whilst the majority of drug vehicle research is focused on the solubility and permeability issues of a drug, significant progress has been made on using vehicles for toxicity reduction. Achieving this can enable safer and more effective use of a potent drug against diseases such as cancer. From a molecular perspective, drugs activate or deactivate biochemical pathways through interactions with cellular macromolecules resulting in toxicity. For newly developed drugs such pathways are not always clearly understood but toxicity endpoints are still required as part of a drug's registration. An understanding of which vehicles could be used to ameliorate the unwanted toxicities of newly developed drugs would be highly desirable to the pharmaceutical industry. In this paper we demonstrate the use of different classifiers as a means to select vehicles best suited to avert a drug's toxic effects when no other information about a drug's characteristics is known. Through analysis of data acquired from the Developmental Therapeutics Program (DTP) we are able to establish a link between a drug's toxicity and vehicle used. We demonstrate that classification and selection of the appropriate vehicle can be made based on the similarity of drug choice.

No description available.

No description available.

No