A value function-based approach for robust surgery planning

Abstract

We present a value function-based formulation for solving a two-stage robust optimization model for surgery-to-OR allocation planning where only upper and lower bounds on surgery duration are known. We solve this formulation using a constraint and column generation algorithm and interpret it as an iterative bi-level interdiction game. We derive valid inequalities based in this interpretation and compare the computational performance of this algorithm with the only other exact solution approach in the literature using the data from an academic medical center. The value function-based formulation yields better or similar quality solutions on difficult problem instances with low or intermediate number of surgeries.