This thesis investigates sepsis management, focusing on diagnostic work-up, antibiotic treatment, and applying data and machine learning to optimize care. Despite decades of research, effective sepsis-specific therapies have yet to be found, limiting the treatment options to supportive therapy and antimicrobials. By implementing a sepsis response team in our hospital’s emergency department, we demonstrated that we could improve the use of the available treatment options. However, there is a clear need for more granular insights to deliver personalized sepsis care. A significant challenge in sepsis research is the highly heterogeneous populations of patients with various infections, comorbidities, and genetic predispositions. In this thesis, we used machine learning techniques such as clustering to find more homogeneous subgroups of sepsis patients who may respond more similarly to specific treatments. We also created prediction models that can give personalized insights to tailor the sepsis management strategy to the individual patient. In addition, we also explored how these machine learning tools can safely and effectively be implemented in clinical practice and how healthcare professionals should be trained to use them appropriately.