Simulation and optimization methods to improve the management of resources and patients in health services: application to emergency departments

• Autores: Marta Cildoz Esquíroz
• Directores de la Tesis: Fermín Mallor Giménez (dir. tes.)
• Lectura: En la Universidad Pública de Navarra ( España ) en 2019
• Idioma: inglés
• Tribunal Calificador de la Tesis: Carmen García Olaverri (presid.) , Inés Marques Proença (secret.) , Elias Willem Hans (voc.)
• Enlaces
  • Tesis en acceso abierto en: Academica-e
• Resumen

  • The aim of this thesis is to contribute to the sustainability of public health services by means of data analysis and through the development and application of Operational Research methods and techniques for modeling and analyzing real planning and management problems generally affecting the public health sector and Emergency Departments (EDs) in particular. The focus of the research is on the development of methods of analysis that will yield practicable solutions to improve the efficiency and quality of patient care and working conditions of the health staff.

    A hospital ED provides medical and/or surgical care to patients arriving in need of immediate attention. The highly stochastic environment of these departments is especially difficult to manage due to the variability of the patient arrival rate, patient severity, and (material and human) health resource requirements. They also have to provide a 24/7 service, where physicians are required to work night, day and weekend shifts, and take on different assignments.

    The research for this thesis covers two types of problem: the improvement of patient flow management and physician shift scheduling. Simulation techniques were selected to model the variable and stochastic environment of the ED. The resulting model includes seasonality in patient arrival patterns by level of severity, and mimics patient pathways through the ED, reflecting the resource consumption (including the medical staff) required for treatment. A guideline is provided for the construction of a mathematical model of the ED designed to overcome some of the shortcomings of oversimplified queuing theory models and capture some important issues that previous simulation models have overlooked.

    The first part of the thesis addresses the problem of patient-to-physician allocation following triage. It offers a proposal for new allocation rules which prove to outperform the common cyclic allocation approach by taking into account a factor usually neglected by patient-flow management policies: i.e., the workload stress experienced by physicians, which is measured in real time using a method proposed and analyzed in this thesis. The stress score is used as the KPI to assess the performance of current patient-flow management policies and as a criterion for designing new ones. This thesis also illustrates the successful implementation of one of the proposed rules, from initial concept to practical application in the hospital. The tested allocation rule outperforms the current cyclic one, as demonstrated by using the simulation model and analysis of the real data gathered during the pilot test.

    The second part of the thesis addresses the physician scheduling problem, which is a combinatorial optimization problem posing particular difficulty when all the constraints and objectives observed in practice are considered. The problem is modeled by means of mathematical programming, and thus cannot be solved in practice by commercial software. This leads to the development of a new solution heuristic. A key feature of this algorithm is the greedy constructive phase, which is guided by solving a linear problem in combination with a memory structure. Initial good solutions are very quickly obtained, but they can be unfeasible in heavily constrained cases. The subsequent improvement phase combines a repair strategy based on variable neighborhood search with network optimization. This is the first proposal for such a strategy. A computational analysis and a real-case solution demonstrate the quality of the solutions and the good behavior of the methodology.

    The research presented in this thesis fulfills the following objectives: • To propose a quantitative framework (based on simulation models and their combination with optimization procedures) for the analysis of problems involved in the dimensioning and assessment of management policies in hospital emergency services.

    • To develop a methodology for the real-time assessment of pending workload stress in physicians.