Thèse Optimisation du Processus d'Analyse de Tumeurs Fraiches H/F Doctorat.Gouv.Fr

Établissement : Université Paris-Saclay GS Sciences de l'ingénierie et des systèmes École doctorale : Sciences Des Systèmes Laboratoire de recherche : Laboratoire Génie Industriel Direction de la thèse : Oualid JOUINI ORCID 000000029498165X Début de la thèse : 2026-12-01 Date limite de candidature : 2026-11-19T23:59:59 Le projet RHU REMISSION vise à améliorer le développement des immunothérapies en oncologie grâce à une caractérisation biologique fine des patients. Il développe notamment PORTRAIT, un programme de profilage moléculaire et immunitaire réalisé à partir de tissus tumoraux frais et de sang, dont l'ambition est de fournir aux oncologues un profil exploitable en moins d'une semaine. L'objectif de cette thèse est de concevoir des modèles et méthodes de recherche opérationnelle permettant de dimensionner, d'organiser et de piloter le processus PORTRAIT lors de son passage à l'échelle, c'est-à-dire d'augmenter les volumes traités tout en maîtrisant les délais, les coûts, l'utilisation des ressources et la qualité de service. Ce processus multi-étapes (transport et stockage des échantillons, dissociation, marquage, cytométrie en flux, dosages multiplex, analyse des données et génération du rapport) est fortement contraint par la périssabilité des échantillons, la variabilité des flux et des temps opératoires, les capacités limitées d'équipements spécialisés et un objectif exigeant de turnaround time. Après une cartographie quantitative du processus, celui-ci sera modélisé par des files d'attente et réseaux de files d'attente, complétés par de la simulation à événements discrets, afin d'identifier les goulets d'étranglement, d'établir des règles de dimensionnement des capacités et de proposer des politiques optimisées de pilotage des flux. La thèse, financée par le projet RHU REMISSION porté par le Prof. Aurélien Marabelle (Institut Gustave Roussy), contribuera directement au process flow diagram et au business case d'industrialisation du processus PORTRAIT. The RHU REMISSION project aims to improve the development of immunotherapies in oncology through the PORTRAIT molecular and immune profiling program, whose ambition is to provide oncologists with an actionable profile within less than one week. Scaling up this program from an academic activity to an industrialized process is a key challenge of the project, addressed by Task T3.6 Upscaling', co-led by the LGI at CentraleSupélec. The PORTRAIT process is a multi-stage diagnostic system subject to strong time constraints and uncertainty: sample perishability, variable flows and processing times, limited capacity of specialized equipment, and stringent quality and traceability requirements. Intra-laboratory operations of this kind remain relatively understudied in the operations research literature. In this context, there is a strong need for decision-support models to size, organize and control the process, so as to increase the volumes processed while controlling lead times, costs, resource utilization and quality of service. The results will directly feed the process flow diagram and the industrialization business case planned in REMISSION. The aim of this PhD thesis is to develop:

(1) a formal, parameterized representation of the PORTRAIT diagnostic production process. The process will first be mapped and quantitatively characterized from field data collected during regular visits to the hospital: activities, flows, routings, human resources, equipment, capacities, processing times, batch operations, precedence constraints and sources of variability. Arrival and service-time distributions will be estimated, and a stochastic model of the system will be developed. Depending on the observed characteristics, the process will be represented by queues and queueing networks, providing analytical results or approximations to identify the main bottlenecks, the sources of variability and the congestion regimes.

(2) methods to size, organize and control the PORTRAIT process as it scales up. Capacity requirements will be determined (number and type of equipment, human resources, opening hours, intermediate buffers and redundancy levels), with the aim of establishing capacity planning rules that guarantee a target service level, in particular in terms of turnaround time, under several demand levels. Flow and capacity optimization models will then be developed, with criteria including total throughput time, compliance with a maximum lead time, work-in-process, resource utilization, investment and operating costs, and robustness to disruptions. A simulation tool will allow industrialization scenarios to be tested before their deployment.
The research will combine field data, analytical modeling and numerical experimentation. A first phase will be devoted to analyzing the actual operation of the process and to estimating arrival and service-time distributions from the available data. Queueing and queueing network models will provide analytical results or approximations useful for capacity sizing and for identifying congestion regimes. Discrete-event simulation will then be used to represent the operational details that cannot be captured analytically. On this basis, flow and capacity optimization models will be developed and evaluated on realistic industrialization scenarios.