Addressing real-world challenges in nurse rostering

The quality of healthcare offerings is of immense value to us all. Hospitals require qualified and devoted nurses to provide the appropriate care to patients. Unfortunately, nurses are a scarce resource, and we face an alarming global shortage. Therefore, providing the nurses with a healthy and enjoyable work environment is essential to the society. Research has shown that round-the-clock shift work can have substantial
negative consequences for employees. When designing rosters, we should focus on reducing the negative effects and allocate the nurses in a way that satisfies both their biological and social needs. The nurse rostering problem is a complex combinatorial optimization problem, and practitioners often find it challenging to generate high-quality solutions without the assistance of decision support tools.

This thesis addresses the nurse rostering problem faced by Danish practitioners. The overall goal has been to develop decision support tools matching their true needs. Thanks to our close collaboration with practitioners in Region Zealand, we have identified practical challenges that researchers have often overlooked. In this thesis, we zoom in on a particular aspect of the nurse rostering problem and provide alternative formulations focused on user understanding and an intuitive optimization objective.

This thesis contributes to the mathematical modeling of the nurse rostering problem and has resulted in four journal articles. The first article includes a mixed integer programming model that generates optimized nurse rosters in Danish hospitals. The following two articles present alternative formulations of the problem, focusing on improving the accuracy of the optimization objective. The final article provides a review of multi-objective optimization methods for personnel rostering problems. We conclude this thesis by putting the research into perspective, reflecting on the past, and providing suggestions for the future.