Novel Metaheuristic Approaches to Nurse Rostering Problems in Belgian Hospitals

Constructing timetables of work for personnel in healthcare institutions is a highly constrained and difficult problem to solve. In this chapter, we will present an overview of our development of the algorithms that underpin a commercial nurse rostering decision support system that is in use in over 40 hospitals in Belgium. As such, we are particularly concerned with the real world regulations and requirements of Belgian institutions. We have concentrated upon short term rostering which involves assigning tasks to nurses in a hospital ward. Of course, the over-riding requirement of our algorithms is to assure a permanent level of care for the patients. However, our approaches are also required to consider administrative requirements, the nurses’ contracts and their personal preferences. In this chapter, we will describe a general model for non-cyclical rostering that copes with a large set of constraints and with the varying objectives that are encountered in practice. The model includes minimum and preferred coverage levels, self-definable (overlapping) qualifications and shift types, different contracts with modifiable constraints and other features. We present a solution framework that applies a modular evaluation function. It also provides many options for initialising and for formulating various objectives and meta-heuristics for searching solutions. Different local search heuristics are applied in different neighbourhoods and we demonstrate the importance of introducing problem specific characteristics into the algorithms. We will conclude by giving an overview and evaluation of the meta-