Discovering Mental Models for the Enhancement of Mental Health Risk Formulation and Clinical Decision Making

Abstract

The uncertain nature of mental health and the complexities in delivering mental healthcare has brought immense pressure on healthcare professionals to use risk assessment and formulation tools that can accommodate the complexities of mental health risk assessment and clinical decision-making processes. The domain of this research is enhancing risk assessment and formulation process in mental health using mental modelling techniques to gaining an understanding of clinical decision making process and clinical workflow. Enhancing risk formulation requires an examination of the clinical decision-making model of the risk formulation tool used and the users’ perceived mental model of the tool based on actual clinical workflow. An elicitation of users’ mental model was carried out with data on users’ interactions with the Galatean Risk and Safety Technology (GRiST); a web risk assessment tool with a view of identifying patterns, behaviours and preferred options of the user which may however not synchronise with the conceptual model of the system causing a mismatch which impacts on performance of the end users. The elicited users’ mental model showed common pattern of data collected and questions ignored when answers are expected. Missing data, incomplete tasks/data and data inconsistencies were common issues identified. This showed the different approach of risk assessment the user has taken; the underlying reasons behind the chosen approach of the user could be lack of understanding of the system and its expectations, confusion arising from the set of questions, non-relevance of the required data or task, time pressure with too many questions, the overriding factor of the clinician’s skills, experience and intuition. This thesis develops a framework aimed at aligning the users’ mental model with the GRIST model is designed to address the shortcomings identified which include omission of data, unanswered questions, incomplete task and or data, non-relevant questions/data, differences in workflow and users’ mental model with GRIST suggested workflow and model is proposed.

Date of Award	Jul 2022
Original language	English
Supervisor	Christopher Buckingham (Supervisor) & Lilit Hakobyan (Supervisor)