Dynamic multicriteria optimization for the nurse scheduling problem

This document addresses the Nurse Scheduling Problem (NSP) and presents a dynamic multicriteria optimization model for its solution considering a predefined time horizon. The purpose
is to maximize the level of "work well-being" of nurses formulated as the minimization of
"aversion" which translates into costs or penalties for certain undesirable work shifts. For this, a
series of criteria are defined to estimate the preference structure of nurses according to the
hospital center specifications by assigning costs for undesirable shift assignments. The proposed
methodology involves developing a heuristic to decompose the global problem into daily
subproblems for which a dynamic algorithm is implemented that considers a cost accumulation
process for all criteria and all nurses. Daily models are dynamically solved by modifying the
coefficients of the well-being function to achieve equity throughout the planning period by
updating and accumulating different averages. This methodology has shown satisfactory results
for scheduling work shifts for doctors, paramedics, security guards, and drivers in numerous
hospital centers in Colombia.