京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第22687号 / 医博第4631号 / 新制||医||1045(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 山下 潤, 教授 木村 剛, 教授 浅野 雅秀 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

Profiling users online and directed online advertising has become a major business with companiessuch as Google and Facebook as frontier companies. Through incidents such as the CambridgeAnalytica scandal, the public has started to take notice of both the positive and the negative sides of thebusiness. The data given to companies with a data driven business model can make the user experiencemore personalized and therefore better. On the other hand the data collected could be seen as privacyreducing and exploitation of users. This study aims to foresee opportunities and new ways to develop adata driven business model which has the user's interests in mind and still remains profitable. Withempirical data through interviews and theories the study will show that data driven business modelshave big potential to be profitable and simultaneously make the user more aware or even make datadelivery in the user’s best interest. The main methods to do this is by implementing privacy dashboards,transparency and moving the pieces in the business model to make the user central in the businessmodel.

This masters thesis deals with analysis of tree structure data. The aim of this thesis is to design and implement a tool for automated detection of relation among samples of read data considering their three structure and node values. Output of the tool is a prescription for automated synthesis of data for testing purposes. The tool is a part of Testos platform developed at FIT BUT.

Simultaneously as the societal trends of urbanization, digitalization and electrification of society are moving at a high speed, the Swedish power grid is undergoing a necessary transition to a renewable energy system. Even though there are difficulties on all grid levels, the lack of capacity in some local grids is among the most present problems and originates from the long lead time of grid expansion as well as the challenges within long-term planning of grids. This thesis aims to improve the understanding of future trends’ impact on grid capacity needs. More specifically, a scenario-based and data driven method, with an accompanying model, is derived to target local capacity challenges. The trends identified to pose impact on the future grid capacity were electrification of different sectors, energy efficiency actions, decentralized energy generation, energy storage solutions, flexibility, smart grids, urbanization and climate. The thesis concludes that the impact of a trend on national level is not simply equal to the impact on a local level. Similarly, a long-term increase of the national electricity consumption does not necessarily worsen local capacity challenges. Furthermore, the developed model in this project shows potential to provide more detailed and accurate information about consumption than currently used methods based on standardized power estimations, which could favor more transparent decision making when dimensioning local grids.

The aim of this paper is to present the possibilities of the usage of advanced analytical tools to optimize decision-making in personnel practice. The literature review part of the thesis deals with the so-called HR analytics, its development, possibilities of its usage, and the methodological framework on which it is based. The next part of the paper deals with the specific application of HR analytics in the field of employee retention according to the methodological framework of CRISP-DM. The last chapter describes in detail the phenomenon of employee turnover, its consequences, and possible explanatory variables. The empirical part of the paper is framed as a quantitative, applied research and deals with voluntary turnover of employees in a particular company-a large Czech bank. Firstly, the statistical-inference part of the research identifies several statistically significant predictors of employee turnover through binary logistic regression-unemployment rate, number of changed teams, time spent in the company, salary and total income, salary growth rate, team size, extraordinary bonus, and gender. Secondly, in the data-science part, several prediction models are compiled, one using binary logistic regression as well and another based on several machine learning techniques. The models are...

Background: Digital technologies are increasingly transforming traditional businesses, and their pervasive impact is leading to a radical restructuring of entire industries. While the significance of generating competitive advantages for businesses utilizing big data analytics is recognized, there is still a lack of consensus of big data analytics influencing strategic decision-making in organisations. As big data and big data analytics become increasingly common, understanding the factors influencing decision-making quality becomes of paramount importance for businesses. Purpose: This thesis investigates how big data and big data analytics affect the operational strategic decision-making processes in enterprises through the theoretical lens of the strategy-as-practice framework. Method: The study follows an abductive research approach by testing a theory (i.e., strategy-aspractice) through the use of a qualitative research design. A single case study of IKEA was conducted to generate the primary data for this thesis. Sampling is carried out internally at IKEA by first identifying the heads of the different departments within the data analysis and from there applying the snowball sampling technique, to increase the number of interviewees and to ensure the collection of enough data for coding. Findings: The findings show that big data analytics has a decisive influence on practitioners. At IKEA, data analysts have become an integral part of the operational strategic decision-making processes and discussions are driven by data and rigor rather than by gut and intuition. In terms of practices, it became apparent that big data analytics has led to a more performance-oriented use of strategic tools and enabling IKEA to make strategic decisions in real-time, which not only increases agility but also mitigates the risk of wrong decisions.

Respiratory motion during PET acquisition degrades image quality. It is mainly the area around the thorax and abdomen which is affected. External devices do provide respiratory gating solutions but are time-consuming to set up on patients and may not always be available. A data-driven gating (DDG) method based on principal component analysis (PCA) was found to provide a reliable respiratory gating signal, discriminating the need for external gating systems with FDG, but it remains to be investigated how well it performs with other PET tracers. The HER2-targeting radiotracer 68Ga-ABY-025 is currently in phase 3 development and is aimed to develop methods to select breast cancer patients that benefit from HER2-targeted treatment. Hence, absolute quantification is important. Respiratory motion correction will be important for improved quantitative accuracy since many patients have metastases in the lower part of the lungs or the liver.  DDG was applied to PET/CT list mode data retrospectively using quiescent period gating. Gated images were then compared to reconstructions without gating with a matched number of coincidences. Two iterative reconstructions were evaluated, TOF OSEM (3 iterations, 16 subsets, and a 5 mm gaussian postprocessing filter) and TOF BSREM β 400. Images were evaluated for standardized uptake value (SUV) changes for well-defined lesions in thorax and abdomen where respiratory motion is prevalent. Respiratory motion was detected in a mean 2.1 bed positions per examination. DDG application resulted in a mean increase of 12.7% in SUVmax for TOF OSEM reconstruction (p=0.0156).

The development of more complex products has increased dependency on virtual/digital models and emphasized the role of simulations as a means of validation before production. This level of dependency on digital models and simulation togetherwith the customization level and continuous requirement change leads to a large number of iterations in each stage of the product development process. This research, studies such group of products that have multidisciplinary, highly iterative, and simulation-driven design processes. It is shown that these high-level technical products, which are commonly outsourced to suppliers, commonly suffer from a long development lead time. The literature points to several research tracks including design automation and data-driven design with possible support. After studying the advantages and disadvantages of each track, a data-driven approachis chosen and studied through two case studies leading to two supporting tools that are expected to improve the development lead time in associated design processes. Feature extraction in CAD as a way to facilitate metamodeling is proposed as the first solution. This support uses the concept of the medial axis to find highly correlated features that can be used in regression models. As for the second supporting tool, an automated CAD script is used to produce a library of images associated with design variants. Dynamic relaxation is used to label each variant with its finite element solution output. Finally, the library is used to train a convolutions neural network that maps screenshots of CAD as input to finite element field answers as output. Both supporting tools can be used to create real-time prediction models in the early conceptual phases of the product development process to explore design space faster and reduce lead time and cost. / Utvecklingen av mer komplexa produkter har ökat beroendet av virtuella/digitala modeller och ökat betydelsen av simuleringar för att validera en produkt inför produktion. Ett stort beroende av digitala modeller och simulering tillsammans med den individuella anpassningen och kontinuerliga kravförändringar leder till ett stort antal iterationer i varje steg i produktutvecklingsprocessen. Forskningen som presenteras i denna avhandling studerar denna typ av produkter som har multidisciplinära, mycket iterativa och simuleringsdrivna designprocesser. Det har visat sig att dessa tekniska produkter på hög nivå, som vanligtvis tillhandahålls av underleverantörer, vanligtvis har en lång ledtid för utveckling. Litteraturstudien pekar på flera forskningsspår, exempelvis designautomation och datadriven design, eventuellt med stöd. Efter att ha studerat fördelarna och nackdelarna med varje spår, väljs det datadrivna tillvägagångssättet och studeras genom två fallstudier som leder till att två stödjande verktyg tas fram. De förväntas förbättra utvecklingsledtiden i tillhörande designprocesser. Feature extraktion i CAD som ett sätt att underlätta metamodellering föreslås som det första verktyget. Detta stöd använder medial axis för att hitta korrelerade features som kan användas i regressionsmodeller. När det gäller det andra stödjande verktyget används ett automatiserat CAD-skript för att producera ett stort bibliotek med bilder som är associerade olika designvarianter. Dynamisk relaxation används för att märka varje variant med dess finita elementlösning. Slutligen används detta bibliotek för att träna ett konvolutionerande neuralt nätverk som kartlägger skärmdumpar av CAD som indata till finita elementfältsvar som utdata. Båda stödverktygen kan användas för att skapa modeller för förutsägelser i realtid i de tidiga konceptuella faserna av produktutvecklingsprocessen för att utforska designrymden snabbare och minska ledtid och kostnader.

A data-driven approach is considered a viable means of dealing with thehigh degree of dynamics caused by the constant changes that occur withinproduction logistics systems. However, there is a dearth of knowledgeregarding the consequences of employing a data-driven approach inproduction logistics in real industrial environments. This thesis aims toextend the existing body of knowledge concerning the opportunities andchallenges of a transition to a data-driven state in relation to productionlogistics through investigating real industrial cases.In addition to reviewing the literature, this thesis aims to answer threeresearch questions. First, it seeks to determine how enabling technologiescontribute to value creation in a data-driven production logistics system.Second, it studies three industrial companies, analyses their productionlogistics flows and compares the tradition approach to a data-drivenapproach by means of discrete event simulation. Third, through interviewswith several experts with different competences who work for the casecompanies, it aims to identify the challenges associated with the transitionto a data-driven approach.The results show that following a systematic and balanced approach totechnology implementation is important with regard to value creation. Thepotential benefits include improved operational performance, improvedvisibility through real-time control and the possibility for dynamicscheduling and planning. The challenges associated with the transition canbe divided into two major categories: organisational and technical.Moreover, the identified challenges can be mapped against each step in theproduction logistics data life-cycle.Among the identified challenges, some represent potentially valuableavenues for future research. Investigating the possibilities for addressingthe data ownership challenge among stakeholders is one such avenue.Additionally, future studies could address the fact that the technologiesrelated to data analytics, such as artificial intelligence, big data andblockchain, lack a large-scale implementation history when compared withtechnologies such as radio frequency identification. Given the limitations ofprior studies, another possible research avenue involves analysing the dataanalytics use cases in more detail within real industrial environments. / Datadrivna metoder betraktas som ett sätt att hantera den höga dynamik som orsakas av ständiga förändringar i industriella system för produktionslogistik. Dock finns det idag begränsad kunskap gällande konsekvenser av att tillämpa datadrivna tillvägagångssätt på produktionslogistik i industriell miljö. Denna avhandling syftar till att utöka den befintliga kunskapen om möjligheter och utmaningar vid övergången till datadriven produktionslogistik genom att utreda verkliga industrifall och genomföra litteraturstudier. Tre forskningsfrågor har formulerats för att nå detta syfte. Först, att utreda hur den möjliggörande tekniken bidrar till värdeskapande i ett datadrivet produktionslogistiksystem. För det andra, att utreda potentiella förbättringar i och med en övergång till datadrivet produktionslogistiksystem, där studier har genomförts på tre industriföretag, deras produktionslogistikflöde samt en jämförelse (genom diskret händelsestyrd simulering) mellan nuläge och börläge. För det tredje, att identifiera utmaningarna vid en övergång till datadrivet produktionslogistiksystem, där flera experter med olika kompetenser har intervjuats i företagen. Resultatet visar att ett systematiskt balanserat tillvägagångssätt för teknikimplementering är viktigt för värdeskapande. Potentiella fördelar inkluderar förbättrad driftsprestanda, förbättrad synlighet genom att ha realtidskontroll och underlätta dynamisk schemaläggning och planering. Övergångsutmaningar är indelade i två huvudkategorier; organisatoriska och tekniska. De identifierade utmaningarna kartläggs mot varje steg i produktionslogistikens livscykel. Bland de identifierade utmaningarna har vissa en särskild potential för framtida forskning. Att undersöka möjligheten att ta itu med utmaningen för dataägande bland intressenter är en av möjligheterna för vidare forskning. Dessutom, teknologier relaterade till dataanalys, såsom AI, big data och block chain har mindre storskalig implementeringshistorik jämfört med annan teknik, såsom RFID. Ett möjligt alternativ för vidare forskning är att analysera användningsfall av dataanalys i mer detalj, givet alla begränsningar som finns inom verklig industriell produktionsmiljö. Nyckelord Produktionslogistik, Data-driven, Smart, Transition, Teknologi, Simulering

Nowadays companies are in a dynamic environment leading to competition in finding new revenue streams to strengthen their positions in their markets by using new technologies to provide capabilitiesto organize resources whilst taking into account changes that can occur in their environment. Therefore, decision making is inevitable to combat uncertainties where taking the optimal action by leveraging concepts and technologies that support decision making such as Business Intelligence (BI)tools and systems could determine a company’s future. Companies can optimize their decision making with BI features like Data-Driven Alerts that sends messages when fluctuations occur within a supervised threshold that reflects the state of business operations. The purpose of this research was to conduct an empirical study on how Swedish companies and enterprises located in different industries apply BI tools and with Data-driven Alerts features for decision making whereby we further studied the characteristics of Data-driven Alerts in terms of usability from the perspectives of different industry professionals through the thematic lens of the Technology acceptance model (TAM) in a qualitative approach. We conducted interviews with professionals from diverse organizations where we applied the Thematic Coding technique on empirical results for further analysis. We found out that by allowing possibilities for users to analyze data in their own preferences for decisions, it will provide managers and leaders with sufficient information needed to empower strategic and tactical decision-making. Despite the emergence of state-of-the-art predictive analytics technologies such as Machine Learning and AI, the literature clearly states that these processes are technical and complex to be comprehended by the decision maker. At the end of the day, prescriptive analytics will end up providing descriptive options being presented to the end user as we move towards automated decision making. This we see as an opportunity for reporting tools and data-driven alerts to be in contemporary symbiotic relationship with advanced analytics in decision making contexts to improve its outcome, quality and user friendliness.