Towards facilitated optimisation

Optimisation modelling in healthcare has addressed a diverse range of challenges inherent to decision-making and supports decision-makers in determining the best solution under a variety of constraints. In contrast, optimisation models addressing planning and service delivery issues in mental healthcare have received limited attention. Mental healthcare services in England are routinely facing issues relative to scarcity of available resources, inequities in their distribution, and inefficiencies in their use. Optimisation modelling has the potential to support decision making and inform the efficient utilisation of scare resources. Mental healthcare services are a combination of several subsystems and partnerships comprising of numerous stakeholders with a diversity of interests. However, in optimisation literature, the lack of stakeholder involvement in the development process of optimisation models is increasingly identified as a missed opportunity impacting the practical applicability of the models and their results. This thesis argues that simulation modelling literature offers alternative modelling approaches that can be adapted to optimisation modelling to address the shortcoming highlighted. In this study, we adapt PartiSim, a multi-methodology framework to support facilitated simulation modelling in healthcare, towards facilitated optimisation modelling and test it using a real case study in mental healthcare. The case study is concerned with a Primary Care Mental Healthcare (PCMH) service that deploys clinicians with different skills to several General Practice (GP) clinics. The service wanted support to help satisfy increasing demand for appointments and explore the possibility of expanding their workforce.

This research puts forward a novel multimethodology framework for participatory optimisation, called PartiOpt. It explores the adaptation and customisation of the and PartiSim framework at each stage of the optimisation modelling lifecycle. The research demonstrates the applicability and relevance of a 'conceptual model' to optimisation modelling, highlighting the potential of facilitated optimisation as a methodology. This thesis argues for the inclusion of conceptual modelling in optimisation when dealing with real world practice-based problems. The thesis proposes an analytics-driven optimisation approach that integrates descriptive, predictive, and prescriptive analytics stages. This approach is utilised to construct a novel multi-skill multi-location optimisation model. By applying the analytics-driven optimisation approach to the case study, previously untapped resource potential is uncovered, leading to the identification of various strategies to improving service efficiency. The successful conceptualisation of an optimisation model and the quantitative decision support requirements that emerged in the initial stages of the study drive the analytics-driven optimisation. Additionally, this research also presents a facilitative approach for stakeholder participation in the validation, experimentation, and implementation of a mathematical optimisation model. Reflecting on the adaptation and subsequent amendments to the modelling stages, the final PartiOpt framework is proposed. It is argued that this framework could reduce the gap between theory and practice for optimisation modelling and offers guidance to optimisation modellers on involving stakeholders in addressing real world problems.