Multi-objective two-stage stochastic programming for adaptive interdisciplinary pain management with piecewise linear network transition models

Abstract

Pain is a major health problem for many people, and pain management is currently innovating because of the opioid crisis in the United States. Existing models optimizing personal adaptive treatment strategies for chronic pain management have only considered one pain outcome. However, most of the pain management centers consider multiple pain outcome measures to identify pain intensity. Consequently, this research uses five pain outcomes. Transition models are represented by piecewise linear networks (PLN). A multi-objective mixed integer linear program (MILP) is developed to optimize treatment strategies for patients based upon on these transition models. A convex quadratic program (QP) is developed to determine weights for multiple levels of multiple pain outcomes that are consistent with surveys submitted by pain management experts. Results show that the MILP that considers multiple pain outcomes yields treatment recommendations with better expected outcomes compared to observed data and to solutions from an optimization model with a single pain outcome objective.