Surgery planning for elective patients: A dedicated heuristic and an effective ALNS. (October 2022)

• Record Type: Journal Article
• Authors: Mezouari, Lahcene; Boufflet, Jean-Paul; Moukrim, Aziz
• Is Part Of: Engineering applications of artificial intelligence. Volume 115(2022)
• Publication Date: 2022-10
• Subjects: Surgery planning -- Project scheduling -- Adaptive heuristic -- Operating room management -- Metaheuristic; Engineering -- Data processing -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Ingénierie -- Informatique -- Périodiques; Intelligence artificielle -- Périodiques; Systèmes experts (Informatique) -- Périodiques; Artificial intelligence; Engineering -- Data processing; Expert systems (Computer science); Periodicals; 620.00285
• DOI: 10.1016/j.engappai.2022.105220
• Languages: English
• ISSNs: 0952-1976

Abstract

Abstract: Hospital organization, the medical concerns of the patient, surgery resources and the horizon to be considered are all elements that contribute to the variety of problems encountered in surgery planning. In this paper, we address the admission planning problem for which surgical interventions of hundreds of elective patients need to be scheduled months before the date of surgery. The health care surgery organization we consider here is based on a shared management of operating rooms and surgeons. The main issue for hospital planners is to schedule all the interventions under resource availability constraints while considering the patients' health priorities. We propose a two-phase 2PSC-EM randomized heuristic that obtains better results on literature benchmark instances. However, for some instances certain interventions are left unscheduled since straightforward heuristic failed to schedule all interventions. We investigated an effective Adaptive Large Neighborhood Search (ALNS) approach. Better results are obtained for each instance, all the patients' interventions are scheduled which had not been done before. The average improvement is about 11.2% and the processing times are shorter than the timeout fixed in the literature, except for one instance for which we succeeded to schedule all of the patients.