Open Access

Milena Grieger

• Machine learning offers significant potential to address challenges in healthcare. This dissertation explores three key contributions of machine learning and evaluates how these applications and methods can improve medical decision-making. Motivated by the pressing issues of escalating costs, nursing shortages, and bureaucratic burdens exacerbated by the Covid-19 pandemic, this study emphasizes the importance of machine learning in assisting clinical staff and enhancing decision-making accuracy in various medical scenarios. The research focuses primarily on supervised learning tasks, with an emphasis on binary classification. It categorizes the three contributions into one application and two methodological advancements within healthcare. The first application investigates the use of analytical and artificial intelligence-based methods in the triage of Covid-19 patients. The two methodological contributions explore the optimization of activation functions and loss functions to improveMachine learning offers significant potential to address challenges in healthcare. This dissertation explores three key contributions of machine learning and evaluates how these applications and methods can improve medical decision-making. Motivated by the pressing issues of escalating costs, nursing shortages, and bureaucratic burdens exacerbated by the Covid-19 pandemic, this study emphasizes the importance of machine learning in assisting clinical staff and enhancing decision-making accuracy in various medical scenarios. The research focuses primarily on supervised learning tasks, with an emphasis on binary classification. It categorizes the three contributions into one application and two methodological advancements within healthcare. The first application investigates the use of analytical and artificial intelligence-based methods in the triage of Covid-19 patients. The two methodological contributions explore the optimization of activation functions and loss functions to improve crucial performance metrics in healthcare, such as sensitivity and area under the curve. The dissertation addresses three critical research questions: the use of loss functions for flexible planning of intensive care unit capacity, the improvement of sensitivity-based binary classification through customized activation functions, and the integration of analytics and machine learning methods to enhance Covid-19 triage while ensuring algorithm explainability. The organization of the dissertation includes an introduction to the topic, an overview of the research contributions, an in-depth discussion of the results, and concluding remarks with directions for future research.…