Pareto-Based Optimization for Multi-objective Nurse Scheduling

In this paper, we propose a search technique for nurse scheduling, which deals with it as a multi-objective problem. For each nurse, we first randomly generate a set of legal shift patterns which satisfy all shift-related hard constraints. We then employ an adaptive heuristic to quickly find a solution with the least number of violations on the coverage-related hard constraint, by assigning one of the available shift patterns to each nurse. Next, we apply a coverage repairing procedure to make the resulting solution feasible, by adding / removing any under-covered / over-covered shifts. Finally, to satisfy the soft constraints (or preferences), we present a simulated annealing based search method with the following two options: one with a weighted-sum evaluation function which encourages moves towards users’ predefined preferences, and another one with a domination-based evaluation function which encourages moves towards a more diversified approximated Pareto set. Computational results demonstrate that the proposed technique is applicable to modern hospital environments.