A Genetic Algorithm-based Method for Creating Impartial Work Schedules for Nurses

The head nurse is generally responsible for creating work schedules for nurses in a nursing unit. However, since versatile constraints, shift requirements, and leave requests are imposed in the problem, the generated schedule usually incurs complaints or criticisms from nurses on its impartiality. This study investigates a complex real nurse scheduling problem and constructs an optimization model for the problem. The model consists of five minimization goals subject to five hard constraints. Then a genetic algorithm-based optimization method is developed for solving the optimization model. The method proposes a dedicated GA coding scheme for the solution to this problem. Based on the scheme the method presents tailored crossover and mutation operators for solution evolution. In the GA method, four selection operation modes are proposed to round in superior chromosomes for the next generation. A personnel data set from a nursing unit of a real hospital is used in numerical tests. In addition, a modified hill climbing greedy method and a well-performed variable depth search method from the literature are implemented to solve the same problem for numerical comparison. Numerical results suggest that a deterministic selection mode outperforms others and our method generates better schedules than other methods on every run.