Patient-based nurse rostering in home care

The monthly nurse rostering is one of the most time consuming planning tasks of our project partner, a leading German home care provider. The planning is currently done manually by experienced senior nurses. However, the planning setup is very complex due to requirements such as legal working time restrictions, availabilities of full and part time nurses, and patient requests with differing requirements on qualification and frequency. The objective is to assign as few different nurses as possible to each patient. Although this continuity is important for quality of care, it is hardly addressed in literature. In this paper, we elaborate a mid-term home care rostering problem (HCR) that fits the requirements of our partner. It implies the assignment of nurses to weekly recurring patient-visit-tours under consideration of the aforementioned continuity of care. We propose different measures of continuity, which are used in five novel MIP formulations, each incorporating all hard planning constraints. We evaluate the models by means of 45 generated instances with realistic features. Further experiments on a real-world instance show that the approach improves continuity of care considerably compared to the manual solution.