No description available.

<p>Predictive modeling in healthcare involves the development of data-driven and computational models which can predict what will happen, be it for a single individual or for an entire system. The adoption of predictive models can guide various stakeholders’ decision-making in the healthcare sector, and consequently improve individual outcomes and the cost-effectiveness of care. With the rapid development in healthcare of big data and the Internet of Things technologies, research in healthcare decision-making has grown in both importance and complexity. One of the complexities facing those who would build predictive models is heterogeneity of patient populations, clinical practices, and intervention outcomes, as well as from diverse health systems. There are many sub-domains in healthcare for which predictive modeling is useful such as disease risk modeling, clinical intelligence, pharmacovigilance, precision medicine, hospitalization process optimization, digital health, and preventive care. In my dissertation, I focus on predictive modeling for applications that fit into three broad and important domains of healthcare, namely clinical practice, public health, and healthcare system. In this dissertation, I present three papers that present a collection of predictive modeling studies to address the challenge of modeling heterogeneity in health care. The first paper presents a decision-tree model to address clinicians’ need to decide among various liver cirrhosis diagnosis strategies. The second paper presents a micro-simulation model to assess the impact on cardiovascular disease (CVD) to help decision makers at government agencies develop cost-effective food policies to prevent cardiovascular diseases, a public-health domain application. The third paper compares a set of data-driven prediction models, the best performing of which is paired together with interpretable machine learning to facilitate the coordination of optimization for hospital-discharged patients choosing skilled nursing facilities. This collection of studies addresses important modeling challenges in specific healthcare domains, and also broadly contribute to research in medical decision-making, public health policy and healthcare systems.</p>

Given that employees are one of the most valuable assets of any organization, losing an employee has a detrimental impact on several aspects of business activities. Loss of competence, deteriorated productivity and increased hiring costs are just a small fraction of the consequences associated with high employee churn. To deal with this issue, organizations within many industries rely on machine learning and predictive analytics to model, predict and understand the cause of employee churn so that appropriate proactive retention strategies can be applied. However, up to this date, the problem of excessive churn prevalent in the healthcare industry has not been addressed. To fill this research gap, this study investigates the applicability of a machine learning-based employee churn prediction model for a Swedish healthcare organization. We start by extracting relevant features from real employee data followed by a comprehensive feature analysis using Recursive Feature Elimination (RFE) method. A wide range of prediction models including traditional classifiers, such as Random Forest, Support Vector Machine and Logistic Regression are then implemented. In addition, we explore the performance of ensemble machine learning model, XGBoost and neural networks, specifically Artificial Neural Network (ANN). The results of this study show superiority of an SVM model with a recall of 94.8% and a ROC-AUC accuracy of 91.1%. Additionally, to understand and identify the main churn contributors, model-agnostic interpretability methods are examined and applied on top of the predictions. The analysis has shown that wellness contribution, employment rate and number of vacations days as well as number of sick day are strong indicators of churn among healthcare employees. / Det sägs ofta att anställda är en verksamhets mest värdefulla tillgång. Att förlora en anställd har därmed ofta skadlig inverkan på flera aspekter av affärsverksamheter. Därtill hör bland annat kompetensförlust, försämrad produktivitet samt ökade anställningskostnader. Dessa täcker endast en bråkdel av konsekvenserna förknippade med en för hög personalomsättningshastighet. För att hantera och förstå hög personalomsättning har många verksamheter och organisationer börjat använda sig av maskininlärning och statistisk analys där de bland annat analyserar beteendedata i syfte att förutsäga personalomsättning samt för att proaktivt skapa en bättre arbetsmiljö där anställda väljer att stanna kvar. Trots att sjukvården är en bransch som präglas av hög personalomsättning finns det i dagsläget inga studier som adresserar detta uppenbara problem med utgångspunkt i maskininlärning. Denna studien undersöker tillämpbarheten av maskininlärningsmodeller för att modellera och förutsäga personalomsättning i en svensk sjukvårdsorganisation. Med utgångspunkt i relevanta variabler från faktisk data på anställda tillämpar vi Recursive Feature Elimination (RFE) som den primära analysmetoden. I nästa steg tillämpar vi flertalet prediktionsmodeller inklusive traditionella klassificerare såsom Random Forest, Support Vector Machine och Logistic Regression. Denna studien utvärderar också hur pass relevanta Neural Networks eller mer specifikt Artificial Neural Networks (ANN) är i syfte att förutse personalomsättning. Slutligen utvärderar vi precisionen av en sammansatt maskininlärningsmodell, Extreme Gradient Boost. Studiens resultat påvisar att SVM är en överlägsen model med 94.8% noggranhet. Resultaten från studien möjliggör även identifiering av variabler som mest bidrar till personalomsättning. Vår analys påvisar att variablerna relaterade till avhopp är friskvårdbidrag, sysselsättningsgrad, antal semesterdagar samt sjuktid är starkt korrelerade med personalomsättning i sjukvården.

This study uses the predictive modeling program Sea-Levels Affecting Marshes Model (SLAMM) to evaluate sea-level rise hazards, such as erosion and inundation, on coastal archaeological sites with a vertical rise of sea level of .98 meters from 2006 to 2100. In total 177 archaeological site locations were collected and georeferenced over GIS outputs maps of wetlands, erosion presence, surface elevation, and accretion. Wetlands data can provide useful information about characteristics of the wetland classes, which make a difference in the ability for coastal archaeological sites to combat sea level rise. Additionally, the study evaluated predicted erosion of archaeological sites by presence or absence of active erosion on a cell-by-cell basis. Elevation map outputs relative to mean tide level allowed for a calculation of individual archaeological site datums to use NOAA tidal databases to identify the potential for their inundation. Accretion maps acquired from the SLAMM run determined the potential for the archaeological site locations to combat rising sea levels and potentially provide protection from wave effects. Results show that the most significant hazard predicted to affect coastal archaeological sites is inundation. Approximately 54% of the total archaeological sites are predicted to be inundated at least half the time by 2100. The hazard of erosion, meanwhile, is expected to affect 33% of all archaeological sites by the end of the century. Although difficult to predict, the study assumes that accretion will not be able to keep pace with sea-level rise. Such findings of hazards prove that SLAMM is a useful tool for predicting potential effects of sea-level rise on coastal archaeological sites. With its ability to customize and as it is complementary, it provides itself not only an economical choice but also one that is adaptable to many scenarios.

Clinical and research practice in the context of Autism rapidly evolved in the last decades. Finer diagnostic procedures, evidence-based models of intervention and higher social inclusivity significantly improved the possibility for autistic children to participate in the fabric of social life. In terms of health best practices, gold-standard procedures still need to be improved, and bridging research and clinical practice still presents several challenges. From the clinical standpoint, the role of process variables, predictors, mechanisms, and timing of change still requires extensive investigation in order to explain response variability and design optimized interventions, tailored to individual needs and maximally effective. Observational techniques represent the elective research methods in child development, especially in clinical contexts, due to their non-invasiveness. However, they still suffer from limited objectivity and poor quantification. Further, their main disadvantage is that they are highly time-consuming and labor-intensive. The aim of this thesis was moving forward to promote translational research in clinical practice of Autism intervention with preschool children. At first, we tried to design and apply quantitative observational techniques to longitudinally study treatment response trajectories during developmental intervention. We tried to characterize different response profiles, and which baseline predictors were able to predict the response over time. Secondly, we investigated mechanisms of change. In particular, we focused on the role of the child-therapist interaction dynamics as a possible active mediator of the process of intervention, especially in the developmental framework that stresses the importance of interpersonal aspects. We also aimed at understanding whether certain time-windows during the intervention were particularly predictive of the response, as well as which specific interaction aspects played a role. Finally, to promote the translational application of observational methods and to improve objective quantification, we proposed and validated an Artificial Intelligence (AI) system to automate data annotation in unconstrained clinical contexts, remaining completely non-invasive and dealing with the specific noisy data that characterize them, for the analysis of the child-therapist acoustic interaction. This effort represents a base building block enabling to employ downstream computational techniques greatly reducing the need for human annotation that usually prevents the application of observational research to large amounts of data . We discuss our findings stressing the importance of assuming a developmental framework in Autism, the key role of the interpersonal experience also in the clinical context, the importance of focusing on trajectories of change and the important need to promote the acquisition of large amounts of quantitative data from the clinical contexts exploiting AI-based systems to assist clinicians, improving objectivity, enabling treatment monitoring, and producing precious data-driven knowledge on treatment efficacy.

No description available.

No description available.

No description available.

<p>  This paper explored the application of three machine learning models focused on predictive motor maintenance. Logistic Regression, Sequential Minimal Optimization (SMO), and NaïveBayes models. A comparative analysis of these models illustrated that while each had an accuracy greater than 95% in this study, the Logistic Regression Model exhibited the most reliable operation.</p>

This thesis investigates the current state of tool wear prediction models in machining, focusing on their limitations in accurately incorporating the complex dynamics of cutting fluids and their industrial applicability. It proposes a comprehensive evaluation framework to classify and evaluate a wide range of models, including empirical, physical, computational, and data-driven models. The study identifies the key limitations and strengths of each model category. It proposes enhancements by integrating advanced fluid dynamics and predictive modeling techniques to improve tool wear predictions' accuracy and industrial applicability. A structured literature review was conducted to investigate and evaluate existing tool wear models and their integration with cutting fluid dynamics. This review included defining search criteria, selecting relevant studies, and assessing their quality and relevance. The study uses thematic analysis and model evaluation frameworks to classify and evaluate the models, leading to the identification of critical limitations and strengths. The literature review and model evaluation findings revealed that empirical models, while simple and quick to implement, showed moderate accuracy and limited fluid dynamics integration. Physical models provided high accuracy in specific conditions but were computationally intensive. Computational models, particularly those using techniques like Finite Element Analysis (FEA) and Computational Fluid Dynamics(CFD), offered detailed insights and high accuracy but required significant computational resources. Data-driven models demonstrated exceptional predictive capabilities and comprehensive fluid dynamics integration but relied heavily on data availability and quality.  The proposed enhancements include introducing non-linear elements into empirical models, incorporating simplified fluid models or empirical correlations into physical models, exploring reduced-order models (ROMs) or surrogate models for computational models, and developing robust data preprocessing and augmentation techniques for data-driven models. These enhancements aim to improve the accuracy and applicability of tool wear models in industrial machining processes, ultimately contributing to more efficient and cost-effective machining operations. The study emphasizes the importance of a systematic and holistic approach to model evaluation and enhancement. Future research should focus on validating these proposed enhancements through empirical studies and real-world applications, ensuring their relevance and robustness in diverse industrial settings. This research offers significant potential to advance tool wear modeling, providing valuable insights for both academia and industry.