Generic approach of multi-level decisions making, contribution to the management of healthcare production system network
French healthcare system confronts the challenges of permanent increase in demand for healthcare, under heavy financial pressure. In the national healthcare strategy, a key focus is to develop a cooperation framework involving all organizations and units. These challenges require healthcare engineering to find efficiency in a more global scale, which means to integrate local optimization problems and decision tools that have generally a high degree of fragmentation in order to contribute to the overall improvement of the system. In this thesis, initiated by a shared unit-dose drug distribution system design project, a generic method was developed to solve the multi-level optimization problem in which interdependent decisions are made at different levels in a hierarchical structure, or at successive stages. The decisions made are often correlated, particularly for decisions in hierarchical topologies that we define by the term "optimal substructure with feedback". The resolution of this problem must be adapted to take into account all implications for correlated decisions. The proposed method is based on the meta-heuristic PSO, it uses a recursive procedure to define the top-down transfer of parameters and the bottom-up feedback of fitness through multiple search spaces, and ensures the consistency of global problem convergence. Our applications and analyzes have shown that this method is generic and is able to provide similar resolution performance and quality compared to the literature references.
KEYWORDS: multi-level optimization, particle swarm optimization, meta-heuristic, generic approach, multi level decision, location-routing problem