Un jumeau numérique hospitalier pour l'optimisation des flux de patients dans les blocs opératoires d'urgence // A hospital digital twin for the optimization of patient flows in emergency operating rooms

Les principales causes de la surcharge des salles d'opération d'urgence (SOU) sont la demande croissante de chirurgies urgentes et la pénurie de personnel médical. Les effets de la surcharge peuvent malheureusement se répercuter sur l'ensemble de l'hôpital, entraînant une mortalité hospitalière plus forte, des séjours plus longs et des coûts plus élevés. Le projet de thèse est motivé par ce contexte, où il devient crucial de repenser l'ingénierie des opérations et l'optimisation des SOU. La thèse sera faite dans le cadre d'une collaboration étroite avec La Pitié Salpêtrière, un des plus grands hôpitaux en France et en Europe.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------------------------------------------------------------------------------

The overcrowding of Emergency Operating Rooms (EORs) has become a critical topic in the healthcare sector (De Simone et al., 2023). Amongst the main causes are the growing demand for non-elective surgeries, the closure of surgery wards in some hospitals to reduce overall system costs, and the unprecedented shortage of medical staff (Lepercq et al., 2021). This situation results in excessive waiting times and reduced access of EOR patients to healthcare services. Unfortunately, the effects of EOR overcrowding can ripple throughout the hospital, resulting in negative system-wide effects such as increasing in-hospital mortality, longer length of stays and higher costs (Santamaria-Acevedo, 2021). Therefore, optimizing EOR patient flows become one of the main priorities of healthcare managers.

The literature on the flow optimization for operating rooms is extensive, but it focuses mainly on elective surgeries (Samudra et al., 2016). Elective surgeries are those scheduled weeks or months in advance, typically involving lower uncertainty and process complexity. The works that incorporate non-elective patients employ simulation-based optimization techniques (Ozcan et al., 2017), as these are well-suited to capture the realistic and complex features of the problem (such as uncertainty in arrivals and activity delays, patient heterogeneity, etc.). However, these works focus on single-stage optimization problems (e.g., patient scheduling for the operating room) without considering the other hospital units involved in delivering patient care. Also, the scheduling problems are often formulated for a single optimization criterion (e.g. costs) and do not account for the inherent dynamics. Henceforth, there is a lack in the literature for studies that consider an integrated optimization for the EOR patient flows within the hospital units, considering healthcare system dynamics and accounting at the same time for multiple KPIs.

Moreover, several researchers believe that a Hospital Digital Twin (HDT) has the potential to facilitate new approaches for the optimization of hospital care activities, particularly by accounting for their dynamics (Elkefi and Asan, 2022; Xames and Topcu, 2024; Pellegrino et al., 2024). However, a digital twin has not yet been developed to support decision-makers in optimizing the delivery of EOR patient care.
The aim of this PhD thesis is to develop:

(1) novel optimization methods for the dynamic scheduling (prioritization) of EOR patients in the different hospital units, namely the EORs, the post-anesthesia care unit (PACU), the intensive care unit (ICU) and the ward. First, independent single-stage dynamic scheduling optimization problems will be formulated for each involved hospital unit using queueing models. Second, meta-heuristic algorithms (such as GAs), integrating the developed single-stage scheduling solutions, will be devised to build near-optimal solutions for the multi-objective global optimization of the EOR patient flows.

(2) a Discrete-Event Simulation (DES) model to evaluate pertinent KPIs of the studied healthcare system. These will include medical KPIs (such as a patient's length of stay in the hospital), social and human science KPIs (such as the work-life balance of hospital staff through stable workloads and minimized overtime), and cost-effectiveness KPIs (such as the cost incurred per patient).

(3) an HDT that allows to generate recommendations on how to prioritize EOR patients in the considered hospital units whenever a new request arrives. The development of HDT will include the definition of an appropriate scheme that integrates the optimization methods and the DES model, its implementation, verification, and validation using Entrepôt de Données de Santé-Assistance Publique Hôpitaux de Paris (EDS APHP) data.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Début de la thèse : 01/10/2025

Autre financement public

ANR

Université Paris-Saclay GS Sciences de l'ingénierie et des systèmes

573 Interfaces : matériaux, systèmes, usages

Candidats intéressés par la recherche opérationnelle, le data science et le management des opérations en santé.
Candidates interested in operations research, data science and operations management for healthcare.