Using agent-based modelling and simulation to model performance measurement in healthcare

Escudero Marin, Paula and Pidd, Michael (2020) Using agent-based modelling and simulation to model performance measurement in healthcare. PhD thesis, Lancaster University.

Text (2020EscuderoPhd)
2020EscuderoPhd.pdf - Published Version
Available under License Creative Commons Attribution Non-commercial No Derivatives.
Download (6MB)

Abstract

One of the priority areas of the UK healthcare system is urgent and emergency care, especially accident and emergency departments (A&E departments). Currently, there is much interest in studying the unintended consequences of the current UK healthcare performance system. Simulation modelling has been proved to be a useful tool for modelling different aspects of the healthcare systems, particularly those related to the performance of A&E departments. Most of the available literature on modelling A&E departments focus on supporting operational decision-making and planning in specific healthcare units to study particular problems such as staff scheduling, resource utilisation, and waiting time issues. That is, most simulation studies focus on analysing how different configurations of healthcare systems affect their performance. However, to our knowledge, few simulation studies focus on explaining how human behaviour affects the performance of the system, and very few have studied how, in turn, performance targets set for A&E departments affect human behaviour in healthcare systems. Some aspects of human behaviour have been incorporated within existing simulation models, though with limitations. In fact, most studies have aimed to study patients’ behaviour, and few have included some aspects of the behaviour of clinical staff. Here we consider how to model clinician behaviour in relation to the performance of A&E departments. This thesis presents an exploratory study of the use of agent-based modelling and simulation (ABMS) and discrete event simulation (DES) to demonstrate how to model clinician behaviour within an A&E department and how that behaviour is related to waiting time performance. Clinical behaviour, incorporated in the simulation models developed here, employs a framework called PECS that assumes that behaviour is influenced by Physical (P), Emotional (E), Cognitive (C) and Social (S) factors. A discussion of the advantages and limitations of the use of ABMS and DES to model such behaviour is included. The findings of this research demonstrate that ABMS is well suited to simulate human behaviour in an A&E department. However, it is not explicitly designed to model processes of complex operational and queue-based systems such as accident and emergency departments. In addition, this research work also demonstrates that DES is an adequate tool for modelling A&E’s processes and patient flows, that can, in fact, incorporate different aspects of human behaviour. Furthermore, the process of modelling human behaviour in DES is complex because, though most DES software allows the representation of reactive behaviour, they make it difficult to model other types of human behaviour The main contributions of this thesis are: 1) a comparison and evaluation of how suitable ABMS and DES are for modelling clinical behaviour, 2) an approach to model the relationship between human behaviour and waiting time performance, considering four aspects of human behaviour (physical, emotional, cognitive and social).