Design of shared unit-dose drug distribution network using multi-level particle swarm optimization

• Published in: Health care management science Vol. 22; no. 2; pp. 304 - 317
• Main Authors: Chen, Linjie, Monteiro, Thibaud, Wang, Tao, Marcon, Eric
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.06.2019; Springer Nature B.V; Springer Verlag
• Subjects: Algorithms; Automation; Business and Management; Computer Science; Drug dosages; Econometrics; France; Health Administration; Health care management; Health Informatics; Hospitals; Humans; Management; Medication Systems, Hospital - economics; Medication Systems, Hospital - organization & administration; Models, Theoretical; Operations Research; Operations Research/Decision Theory; Optimization; Organization theory; Pharmacy Service, Hospital - organization & administration; Production planning; France; Facility location; Multi-level optimization; Process improvement; Patient safety; Meta-heuristic; Logistic management; Particle swarm optimization; Capacitated location routing problem
• ISSN: 1386-9620; 1572-9389; 1572-9389
• DOI: 10.1007/s10729-018-9438-6

Unit-dose drug distribution systems provide optimal choices in terms of medication security and efficiency for organizing the drug-use process in large hospitals. As small hospitals have to share such automatic systems for economic reasons, the structure of their logistic organization becomes a very sensitive issue. In the research reported here, we develop a generalized multi-level optimization method – multi-level particle swarm optimization (MLPSO) – to design a shared unit-dose drug distribution network. Structurally, the problem studied can be considered as a type of capacitated location-routing problem (CLRP) with new constraints related to specific production planning. This kind of problem implies that a multi-level optimization should be performed in order to minimize logistic operating costs. Our results show that with the proposed algorithm, a more suitable modeling framework, as well as computational time savings and better optimization performance are obtained than that reported in the literature on this subject.