A mixed integer programming model for Bed planning considering stochastic length of stay

In this paper, we study hospital bed planning problem for a gynecological ward in China. We first proposed a mixed integer programming (MIP) model considering deterministic length of stay (LOS) to assign patients to available beds. Since statistical analysis of empirical hospitalization data reveals that variance of inpatient´s LOS is significant. Therefore, based on the deterministic LOS model we further proposed a MIP model taking stochastic LOS into consideration. Both models could be solved by standard linear programming solver CPLEX. Numerical experiments of bed assignment results comparison between the two proposed models are presented. The result shows stochastic LOS model can generate better bed planning solutions with lower potential schedule conflict cost.