Knowledge engineering for mental-health risk assessment and decision support

Mental-health risk assessment practice in the UK is mainly paper-based, with little standardisation in the tools that are used across the Services. The tools that are available tend to rely on minimal sets of items and unsophisticated scoring methods to identify at-risk individuals. This means the reasoning by which an outcome has been determined remains uncertain. Consequently, there is little provision for: including the patient as an active party in the assessment process, identifying underlying causes of risk, and eecting shared decision-making. This thesis develops a tool-chain for the formulation and deployment of a computerised clinical decision support system for mental-health risk assessment. The resultant tool, GRiST, will be based on consensual domain expert knowledge that will be validated as part of the research, and will incorporate a proven psychological model of classication for risk computation. GRiST will have an ambitious remit of being a platform that can be used over the Internet, by both the clinician and the layperson, in multiple settings, and in the assessment of patients with varying demographics. Flexibility will therefore be a guiding principle in the development of the platform, to the extent that GRiST will present an assessment environment that is tailored to the circumstances in which it nds itself. XML and XSLT will be the key technologies that help deliver this exibility.

Defining an OWL ontology that could be used to integrate mental-health risk information within a decision support system

In the field of mental health risk assessment, there is no standardisation between the data used in different systems. As a first step towards the possible interchange of data between assessment tools, an ontology has been constructed for a particular one, GRiST (Galatean Risk Screening Tool). We briefly introduce GRiST and its data structures, then describe the ontology and the benefits that have already been realised from the construction process. For example, the ontology has been used to check the consistency of the various trees used in the model. We then consider potential uses in integration of data from other sources. © 2009 IEEE.

Using XML and XSLT for flexible elicitation of mental-health risk knowledge

Current tools for assessing risks associated with mental-health problems require assessors to make high-level judgements based on clinical experience. This paper describes how new technologies can enhance qualitative research methods to identify lower-level cues underlying these judgements, which can be collected by people without a specialist mental-health background. Content analysis of interviews with 46 multidisciplinary mental-health experts exposed the cues and their interrelationships, which were represented by a mind map using software that stores maps as XML. All 46 mind maps were integrated into a single XML knowledge structure and analysed by a Lisp program to generate quantitative information about the numbers of experts associated with each part of it. The knowledge was refined by the experts, using software developed in Flash to record their collective views within the XML itself. These views specified how the XML should be transformed by XSLT, a technology for rendering XML, which resulted in a validated hierarchical knowledge structure associating patient cues with risks. Changing knowledge elicitation requirements were accommodated by flexible transformations of XML data using XSLT, which also facilitated generation of multiple data-gathering tools suiting different assessment circumstances and levels of mental-health knowledge. © 2007 Informa UK Ltd All rights reserved.

Investigating mental health risk assessment in primary care and the potential role of a structured decision support tool, GRiST

Aim: To explore current risk assessment processes in general practice and Improving Access to Psychological Therapies (IAPT) services, and to consider whether the Galatean Risk and Safety Tool (GRiST) can help support improved patient care. Background: Much has been written about risk assessment practice in secondary mental health care, but little is known about how it is undertaken at the beginning of patients' care pathways, within general practice and IAPT services. Methods: Interviews with eight general practice and eight IAPT clinicians from two primary care trusts in the West Midlands, UK, and eight service users from the same region. Interviews explored current practice and participants' views and experiences of mental health risk assessment. Two focus groups were also carried out, one with general practice and one with IAPT clinicians, to review interview findings and to elicit views about GRiST from a demonstration of its functionality. Data were analysed using thematic analysis. Findings Variable approaches to mental health risk assessment were observed. Clinicians were anxious that important risk information was being missed, and risk communication was undermined. Patients felt uninvolved in the process, and both clinicians and patients expressed anxiety about risk assessment skills. Clinicians were positive about the potential for GRiST to provide solutions to these problems. Conclusions: A more structured and systematic approach to risk assessment in general practice and IAPT services is needed, to ensure important risk information is captured and communicated across the care pathway. GRiST has the functionality to support this aspect of practice.

Handling varying amounts of missing data when classifying mental-health risk levels

One of the main challenges of classifying clinical data is determining how to handle missing features. Most research favours imputing of missing values or neglecting records that include missing data, both of which can degrade accuracy when missing values exceed a certain level. In this research we propose a methodology to handle data sets with a large percentage of missing values and with high variability in which particular data are missing. Feature selection is effected by picking variables sequentially in order of maximum correlation with the dependent variable and minimum correlation with variables already selected. Classification models are generated individually for each test case based on its particular feature set and the matching data values available in the training population. The method was applied to real patients' anonymous mental-health data where the task was to predict the suicide risk judgement clinicians would give for each patient's data, with eleven possible outcome classes: zero to ten, representing no risk to maximum risk. The results compare favourably with alternative methods and have the advantage of ensuring explanations of risk are based only on the data given, not imputed data. This is important for clinical decision support systems using human expertise for modelling and explaining predictions.

Integrating service user and practitioner expertise within a web-based system for collaborative mental-health risk and safety management

Objectives: To develop a decision support system (DSS), myGRaCE, that integrates service user (SU) and practitioner expertise about mental health and associated risks of suicide, self-harm, harm to others, self-neglect, and vulnerability. The intention is to help SUs assess and manage their own mental health collaboratively with practitioners. Methods: An iterative process involving interviews, focus groups, and agile software development with 115 SUs, to elicit and implement myGRaCE requirements. Results: Findings highlight shared understanding of mental health risk between SUs and practitioners that can be integrated within a single model. However, important differences were revealed in SUs' preferred process of assessing risks and safety, which are reflected in the distinctive interface, navigation, tool functionality and language developed for myGRaCE. A challenge was how to provide flexible access without overwhelming and confusing users. Conclusion: The methods show that practitioner expertise can be reformulated in a format that simultaneously captures SU expertise, to provide a tool highly valued by SUs. A stepped process adds necessary structure to the assessment, each step with its own feedback and guidance. Practice Implications: The GRiST web-based DSS (www.egrist.org) links and integrates myGRaCE self-assessments with GRiST practitioner assessments for supporting collaborative and self-managed healthcare.

Developing a probabilistic graphical structure from a model of mental-health clinical risk expertise

This paper explores the process of developing a principled approach for translating a model of mental-health risk expertise into a probabilistic graphical structure. The Galatean Risk Screening Tool [1] is a psychological model for mental health risk assessment based on fuzzy sets. This paper details how the knowledge encapsulated in the psychological model was used to develop the structure of the probability graph by exploiting the semantics of the clinical expertise. These semantics are formalised by a detailed specification for an XML structure used to represent the expertise. The component parts were then mapped to equivalent probabilistic graphical structures such as Bayesian Belief Nets and Markov Random Fields to produce a composite chain graph that provides a probabilistic classification of risk expertise to complement the expert clinical judgements. © Springer-Verlag 2010.

"Crack down on the celebrity junkies": does media coverage of celebrity drug use pose a risk to young people?

This study analysed news media content to examine the role played by celebrity drug use in young people's perceptions of drug use. We know that young people have access to discourses of drug use through music and other media which may emphasise short term gains (of pleasure or sexual success) over longer term health and social problems. This study goes beyond a simple modelling approach by using Media Framing Analysis (MFA) to take an in-depth look at the messages themselves and how they are 'framed'. New stories about Amy Winehouse's drug use were used and we conducted focus groups with young people asking them questions about drugs, celebrity and the media. Frames identified include: 'troubled genius', 'losing patience' and 'glamorization or gritty realism'. Initially, the press championed Winehouse's musical talent but soon began to tire of her recklessness; the participants tended to be unimpressed with Winehouse's drug use, characterising her as a promising artist who had 'gone off the rails'. Young people were far more critical of Winehouse than might be expected, demonstrating that concerns about the influence of celebrity drug use and its impact on future health risk behaviour among young people may have been over-simplified and exaggerated. This study illustrates the need to understand young people and their frames of reference within popular culture when designing drug awareness information relevant to them. Furthermore, it indicates that critical media skills analysis may contribute to health risk education programmes related to drug use.

Development of a probabilistic graphical structure from a model of mental health clinical expertise

This thesis explores the process of developing a principled approach for translating a model of mental-health risk expertise into a probabilistic graphical structure. Probabilistic graphical structures can be a combination of graph and probability theory that provide numerous advantages when it comes to the representation of domains involving uncertainty, domains such as the mental health domain. In this thesis the advantages that probabilistic graphical structures offer in representing such domains is built on. The Galatean Risk Screening Tool (GRiST) is a psychological model for mental health risk assessment based on fuzzy sets. In this thesis the knowledge encapsulated in the psychological model was used to develop the structure of the probability graph by exploiting the semantics of the clinical expertise. This thesis describes how a chain graph can be developed from the psychological model to provide a probabilistic evaluation of risk that complements the one generated by GRiST’s clinical expertise by the decomposing of the GRiST knowledge structure in component parts, which were in turned mapped into equivalent probabilistic graphical structures such as Bayesian Belief Nets and Markov Random Fields to produce a composite chain graph that provides a probabilistic classification of risk expertise to complement the expert clinical judgements

Self-affirmation before exposure to health communications promotes intentions and health behavior change by increasing anticipated regret

Health-risk information can elicit negative emotions like anticipated regret that may positively affect health persuasion. The beneficial impact of such emotions is undermined when target audiences respond defensively to the threatening information. We tested whether self-affirming (reflecting on cherished attributes) before message exposure can be used as strategy to enhance the experience of anticipated regret. Women were self-affirmed or not before exposure to a message promoting fruit and vegetable consumption. Self-affirmation increased anticipated regret and intentions reported following message exposure and consumption in the week after the intervention; regret mediated the affirmation effect on intentions. Moreover, results suggest that anticipated regret and intentions are serial mediators linking self-affirmation and behavior. By demonstrating the mediating role of anticipated regret, we provide insights into how self-affirmation may promote healthy intentions and behavior following health message exposure. Self-affirmation techniques could thus potentially be used to increase the effectiveness of health communication efforts.

Advocating alcohol abstinence to pregnant women: some observations about British policy

In 2007, the English Department of Health (DH) issued advice stating 'pregnant woman' and 'those trying to conceive' should abstain from drinking alcohol. As others have noted, this advice was issued despite their being no new evidence about the deleterious effects of low levels of alcohol consumption. In this paper, we argue this development is significant for the social construction of 'risk', since in advocating abstinence without an evidence base for this advice, policy makers formalise a connection between uncertainty and danger. We suggest this development has important implications, most obviously for pregnant women, certainly impacting on the nature of the advice they will now receive and likely more generally on their experience of the transition to motherhood. We suggest it has wider implications for individuals' experience also, as policy makers appear to be advocating the same approach to risk to non-pregnant people. Further, it suggests a noteworthy formalisation of a new definition of risk, which should be debated far more extensively, as it matters for the future development of health policy.

Accountability, organisational learning and risks to patient safety in England: conflict or compromise?

This article examines the current risk regulation regime, within the English National Health Service (NHS), by investigating the two, sometimes conflicting, approaches to risk embodied within the field of policies towards patient safety. The first approach focuses on promoting accountability and is built on legal principles surrounding negligence and competence. The second approach focuses on promoting learning from previous mistakes and near-misses, and is built on the development of a ‘safety culture’. Previous work has drawn attention to problems associated with risk-based regulation when faced with the dual imperatives of accountability and organisational learning. The article develops this by considering whether the NHS patient safety regime demonstrates the coexistence of two different risk regulation regimes, or merely one regime with contradictory elements. It uses the heuristic device of ‘institutional logics’ to examine the coexistence of and interrelationship between ‘organisational learning’ and ‘accountability’ logics driving risk regulation in health care.

The multiple pheromone ant clustering algorithm and its application to real world domains

The Multiple Pheromone Ant Clustering Algorithm (MPACA) models the collective behaviour of ants to find clusters in data and to assign objects to the most appropriate class. It is an ant colony optimisation approach that uses pheromones to mark paths linking objects that are similar and potentially members of the same cluster or class. Its novelty is in the way it uses separate pheromones for each descriptive attribute of the object rather than a single pheromone representing the whole object. Ants that encounter other ants frequently enough can combine the attribute values they are detecting, which enables the MPACA to learn influential variable interactions. This paper applies the model to real-world data from two domains. One is logistics, focusing on resource allocation rather than the more traditional vehicle-routing problem. The other is mental-health risk assessment. The task for the MPACA in each domain was to predict class membership where the classes for the logistics domain were the levels of demand on haulage company resources and the mental-health classes were levels of suicide risk. Results on these noisy real-world data were promising, demonstrating the ability of the MPACA to find patterns in the data with accuracy comparable to more traditional linear regression models. © 2013 Polish Information Processing Society.

Psychological cue use and implications for a clinical decision support system

Effective clinical decision making depends upon identifying possible outcomes for a patient, selecting relevant cues, and processing the cues to arrive at accurate judgements of each outcome's probability of occurrence. These activities can be considered as classification tasks. This paper describes a new model of psychological classification that explains how people use cues to determine class or outcome likelihoods. It proposes that clinicians respond to conditional probabilities of outcomes given cues and that these probabilities compete with each other for influence on classification. The model explains why people appear to respond to base rates inappropriately, thereby overestimating the occurrence of rare categories, and a clinical example is provided for predicting suicide risk. The model makes an effective representation for expert clinical judgements and its psychological validity enables it to generate explanations in a form that is comprehensible to clinicians. It is a strong candidate for incorporation within a decision support system for mental-health risk assessment, where it can link with statistical and pattern recognition tools applied to a database of patients. The symbiotic combination of empirical evidence and clinical expertise can provide an important web-based resource for risk assessment, including multi-disciplinary education and training. © 2002 Informa UK Ltd All rights reserved.

Growing better brains? Pregnancy and neuroscience discourses in English social and welfare policies

In recent years, English welfare and health policy has started to include pregnancy within the foundation stage of child development. The foetus is also increasingly designated as ‘at risk’ from pregnant women. In this article, we draw on an analysis of a purposive sample of English social and welfare policies and closely related advocacy documents to trace the emergence of neuroscientific claims-making in relation to the family. In this article, we show that a specific deterministic understanding of the developing brain that only has a loose relationship with current scientific evidence is an important component in these changes. We examine the ways in which pregnancy is situated in these debates. In these debates, maternal stress is identified as a risk to the foetus; however, the selective concern with women living in disadvantage undermines biological claims. The policy claim of neurological ‘critical windows’ also seems to be influenced by social concerns. Hence, these emerging concerns over the foetus’ developing brain seem to be situated within the gendered history of policing women’s pregnant bodies rather than acting on new insights from scientific discoveries. By situating these developments within the broader framework of risk consciousness, we can link these changes to wider understandings of the ‘at risk’ child and intensified surveillance over family life.