Abstract
Demographic pressures from an ageing population, budgetary constraints and organisational restructuring increase, amongst others, the demand for better planning and management of health services. However, since health care systems are complex and stochastic, simple approaches for planning and managing are expected to be inadequate. These simple approaches fail to describe sufficiently the interaction between different groups of patients and its effect on the whole system. Patients flow through the different stages of care in different streams and as a result the temporal characteristics of patient populations are rarely homogenous. In this thesis, we put forward a decision support framework that is specifically designed to identify the different streams of patient flow and to investigate their interaction from a behavioural perspective. The richness of the available data dictates the use of data warehousing and On-Line Analytical Processing (OLAP) for data analysis and pre-processing; the complex nature of health care systems suggests the use of simulation for the provision of a decision model.
More specifically, the main contribution of this work is threefold. First, an already established deterministic model of patient flow is adapted to a queuing model and discrete event simulation is employed to evaluate it numerically. Second, a comprehensive framework for analysing hospital length of stay and bed occupancy in an OLAP-enabled data warehouse environment is described. Third, the application of data warehousing and OLAP techniques to analysing the output of data intensive simulation experiments is demonstrated. This is, to my knowledge, the first effort to take advantage of the OLAP capabilities in analysing output from discrete event simulation models. The application of the framework is then demonstrated by applying it to nationwide and hospital-wide datasets and in particular to model specific problematic situations such as the winter bed crisis.
In summary, in this thesis we propose the incorporation
of discrete event simulation modelling and data warehousing techniques into
a decision support framework for modelling the flow of patients through hospitals
and health care systems. This framework can be easily applied to model different
levels of health care operations. The scalability of the individual modelling
components make this framework unique in its kind