Abstract
Healthcare organisations increasingly use simulation and modelling techniques to analyse their procedures and policies. Modelling activities attempt to help meet the challenges, constraints and requirements for efficiency encountered in the modern healthcare environment. A variety of techniques are used, often applied in different roles and by different functions in the organisation. Recent research has investigated the benefits of considering multiple approaches in the analysis of problems. This chapter briefly introduces the use of simulation and modelling in healthcare and the factors driving the increasingly widespread use of these techniques. Simple examples show how individual methods may be applied to model healthcare problems. The recent emergence of multi method approaches to modelling is examined and, focusing specifically on healthcare, examples of how these new ideas may also be applied in healthcare modelling are presented. Finally the challenges to implementing such new approaches effectively in a healthcare environment are discussed.
TopThe Changing Profile Of Modelling In Healthcare
Using abstract models to understand the behaviour of a subject is common in disciplines such as economics and engineering, as a result where these disciplines overlap with healthcare, models are often used. Increasingly however models are also being applied to issues of process, organisation and cost in healthcare using a wide range of techniques. Briggs & Claxton et al. (2006) presents some approaches for evaluating costs including decision trees, Markov modelling and simulation techniques. Morris & Devlin et al. (2007) discusses economic and statistical models. Fone & Hollinghurst et al. (2003) provides an extensive survey of healthcare modelling studies, particularly where simulation methods are used, and identifies four main areas for their application; hospital scheduling and organisation, infections and communicable diseases, costs and economic evaluation and screening.
Key Terms in this Chapter
Discrete Event Simulation: A simulation based form of modelling in which patterns of events in the problem are recreated so that the timing and resource implications can be examined. The events generated usually include the arrival and departure of entities from the system or one of its sub processes. Timing and quantities in the model are typically generated by implementing appropriate stochastic distributions. Law & Kelton (2000) provides a comprehensive introduction to this form of modelling.
Organisational Modelling: Modelling undertaken to understand or analyse the effects of the structure, processes, and policies of an organisation. The role of organisational modelling is typically to understand complex observed behaviour, to prototype new policies and configurations or to document existing processes. In healthcare organisations such studies are increasingly used to investigate costs, patient flows and the utilisation of scarce resources. These studies may be contrasted with problems where organisational issues have negligible or no impact; in healthcare these may include anatomical and epidemiological modelling.
Modelling Resource Load: A concept used to compare the relative resource costs of modelling studies using different methods. For small scale studies the difference in resources required may be negligible, however in large studies this factor may be used to determine the most appropriate choice of modelling method. The differences in between methods in problem conceptualisation and model implementation determine the resources necessary to implement the model. In assessing the scale three key factors are relevant; the scope, or size of the problem; the required amount of detail per entity; the time scale considered by the study.
Modelling Resource Load: A concept used to compare the relative resource costs of modelling studies using different methods. For small scale studies the difference in resources required may be negligible, however in large studies this factor may be used to determine the most appropriate choice of modelling method. The differences in between methods in problem conceptualisation and model implementation determine the resources necessary to implement the model. In assessing the scale three key factors are relevant; the scope, or size of the problem; the required amount of detail per entity; the time scale considered by the study.
Markov Modelling: A form of modelling based on stochastic processes where the discrete states of a problem and the possible transitions between them are analysed. Systems represented as a network of states with paths between nodes weighted according to their probability of their occurrence. Paths and cycles in the system can be analysed mathematically to determine the likelihood of overall outcomes. The Markov property requires that the probability of transition between two states is dependant only on the current state and the problem must be formulated accordingly.
Agent Based Modelling: A simulation based modelling approach, where the problem is represented the using software entities with some degree of autonomy. The behaviour observed in the problem is recreated through the interaction of the entities, known as agents, which are typically modelled using data objects with encapsulated behaviour. Models may exhibit complex emergent behaviour even where the rules governing individual behaviour are relatively simple. They are sometimes considered to combine the analytical features of both system dynamics and discrete event modelling.
Econometric Modelling: A form of modelling based on the integration of statistical measurements of problem variables with economic theory. Models are based on empirical data collected from the problem and variables may be treated as deterministic or stochastic. A wide variety of statistical techniques are used to measure the quantities in the problem and assess their underlying distributions. These may be used to test hypotheses about the problem and make predictions about future behaviour or the likely effects of policy changes.
Multi Method Modelling: An emerging area in modelling methodology where techniques from different modelling disciples are combined inorder to analyse a single problem. The premise of this approach is that different methods have different features which can, in some cases, be used to support a more effective study. There is particular interest in applying this approach to healthcare problems because of the potential to extend existing model, integrate different stakeholder views and overcome technical limitations in some studies.
Multi Method Modelling: An emerging area in modelling methodology where techniques from different modelling disciples are combined inorder to analyse a single problem. The premise of this approach is that different methods have different features which can, in some cases, be used to support a more effective study. There is particular interest in applying this approach to healthcare problems because of the potential to extend existing model, integrate different stakeholder views and overcome technical limitations in some studies.
Markov Modelling: A form of modelling based on stochastic processes where the discrete states of a problem and the possible transitions between them are analysed. Systems represented as a network of states with paths between nodes weighted according to their probability of their occurrence. Paths and cycles in the system can be analysed mathematically to determine the likelihood of overall outcomes. The Markov property requires that the probability of transition between two states is dependant only on the current state and the problem must be formulated accordingly.
System Dynamics Modelling: A systems-oriented simulation based modelling approach first proposed in Forrester (1961). Models are based on the causal structure of the problem including the perceptions of the actors. Two levels of modelling are possible; Qualitative modelling using influence diagrams and causal loop analysis or quantitative modelling using Stock-Flow diagrams and computer simulation.
Organisational Modelling: Modelling undertaken to understand or analyse the effects of the structure, processes, and policies of an organisation. The role of organisational modelling is typically to understand complex observed behaviour, to prototype new policies and configurations or to document existing processes. In healthcare organisations such studies are increasingly used to investigate costs, patient flows and the utilisation of scarce resources. These studies may be contrasted with problems where organisational issues have negligible or no impact; in healthcare these may include anatomical and epidemiological modelling.
Agent Based Modelling: A simulation based modelling approach, where the problem is represented the using software entities with some degree of autonomy. The behaviour observed in the problem is recreated through the interaction of the entities, known as agents, which are typically modelled using data objects with encapsulated behaviour. Models may exhibit complex emergent behaviour even where the rules governing individual behaviour are relatively simple. They are sometimes considered to combine the analytical features of both system dynamics and discrete event modelling.
Discrete Event Simulation: A simulation based form of modelling in which patterns of events in the problem are recreated so that the timing and resource implications can be examined. The events generated usually include the arrival and departure of entities from the system or one of its sub processes. Timing and quantities in the model are typically generated by implementing appropriate stochastic distributions. Law & Kelton (2000) provides a comprehensive introduction to this form of modelling.
Econometric Modelling: A form of modelling based on the integration of statistical measurements of problem variables with economic theory. Models are based on empirical data collected from the problem and variables may be treated as deterministic or stochastic. A wide variety of statistical techniques are used to measure the quantities in the problem and assess their underlying distributions. These may be used to test hypotheses about the problem and make predictions about future behaviour or the likely effects of policy changes.
System Dynamics Modelling: A systems-oriented simulation based modelling approach first proposed in Forrester (1961). Models are based on the causal structure of the problem including the perceptions of the actors. Two levels of modelling are possible; Qualitative modelling using influence diagrams and causal loop analysis or quantitative modelling using Stock-Flow diagrams and computer simulation.