Health Care Simulation-Optimization

The operations research program of the University of Auckland includes some high quality and very successful simulations for the health care sector. These simulations include the transportation of patients, rostering of employees, scheduling of tasks as well as allocation of resources and patient flows. Although some of these simulations already help Auckland Hospital to improve its quality of patient care as well as decrease its operation costs, others need some finalization work in order to be used in real world applications. The desired project should build on this foundation and includes two main tasks which are briefly described in the sequel. The first task is to finalize the work previously done at the Engineering Department of the University of Auckland. As a leading research facility in operations research, including simulation, they have launched many interesting and state-of-the-art simulation projects. However, some require finalization steps to be used more effectively in heterogeneous environments. To ensure that concepts, algorithms and methods can be used on a variety of case studies, applications and further research topics, one of our main requirement on the project is to conserve the findings and ideas in a conceptual framework as well as in a software library. Thereby, simulation can be more easily, quickly and cost efficient used as a decision tool within various fields of health care and so contributes to a higher quality of patient care. The second task is the integration of meta-heuristics, in particular genetic algorithms, in simulation concepts. By doing so, the main question must be how are we able to use operations research to improve the quality of health care in Europe which is based on the developed simulations and models. While simulation helps decision makers to understand and analyse behaviours of systems, optimization provides additional theory and methods to improve processes in terms of both quality of patient care and efficiency. Unless some attempts have been undertaken to integrate optimization concepts in health care in the past, most of them lack in scientific methodology and discussion. Therefore, an exciting new area within operations research can be created and studied. Since the Department of Engineering Science at the University of Auckland is a leading research institution for operations research outside the U.S., the training and learning prospects, as well as chances to gain knowledge which can be transferred to Europe, are excellent. In addition, the focus on making concepts and techniques more general leads to a variety of possibilities to use these gained skills to improve the Austrian or even the European health care system.

Operations research methodologies, such as modelling, simulation and optimization, have been successfully used in many industries to analyze, design and improve systems. However, the number of successful applications in the health care sector is relatively small compared to other industries, such as manufacturing, defense sector or logistics. Within this project possible explanations for this deficit have been identified and analyzed. A detailed investigation of health care systems showed that they consist of highly dynamic and variable decision strategies for resource allocation and task dispatching compared to traditional domains. As a result techniques and methods motivated by those traditional application domains are not capable of capturing the dynamic nature and flexibility of health care operations adequately. Hence, by the design of new modeling frameworks and simulation tools, that break with rigid assumptions of conventional paradigms and are motivated by the requirements of the health care industry, the project contributes to the acceptance and success of modeling simulation in this domain. The limited re-use of models and simulation solutions over the course of multiple studies, especially in the health care sector, is a commonly addressed challenge for researchers. The methods and frameworks developed in this project provide welldefined guidelines to build and document conceptual models. Such a structured definition and documentation of conceptual models is an inevitable requirement for the re-use of models and simulation solutions. Further, commonly used building blocks of health care models were identified and included in the previously mentioned guidelines. The resulting generic toolkit enables modelers to adapt and combine predefined elements to quickly build problem specific conceptual models. Thereby, development time can be reduced and conceptual re-use is enhanced. Based on the conceptual results, a software library was implemented that transports the conceptual re-use to the implementation phase of a simulation study. By providing generic, modular, and adaptable implementations of conceptual building blocks, development effort for simulation solutions can be significantly reduced. The most common way to combine simulation models and optimization methods employs models for the evaluation of decisions made by an optimization algorithm as an objective function. The structure and design of the conceptual framework, as well as the software library, allows the integration of optimization methods within models. Thereby, the continuous use of automated optimization procedures in real systems can be imitated and their performance can be evaluated accurately. The newly developed methods and tools have already been used for two case studies in the course of this project. The first study dealt with the analysis of the care-process breast cancer patients traverse and the evaluation of the potential improvement resulting from accelerations of single process steps. The second study included the design and implementation of optimization algorithms for patient transit dispatching. The goal this study was to minimize the number of late transports by the computation of optimal task sequences for single and combined resources. Previously designed algorithms do not account for the additional requirements often inherent to the transportation of patients. Due to the variety of health conditions patients may display during transportation, multiple resource configurations may be required to perform transits. Algorithms tailored to this specification were designed, implemented and evaluated by the use of a simulation model. Both studies highlighted the strengths of the newly developed methods. Hence, institutions participating in this project declared those methods as their new standard to develop simulation models.