Predictive Risk Modeling

The financial services industry is undergoing a transformative shift due to the advent of Artificial Intelligence (AI), particularly in the realm of predictive risk management. This paper explores the integration of AI technologies in identifying, assessing, and mitigating risks within financial institutions. By leveraging machine learning algorithms, big data analytics, and realtime processing capabilities, financial organizations can anticipate potential risks, enhance decision-making processes, and improve operational efficiency. This study examines the current landscape of AI applications in risk management, highlights case studies of successful implementations, and discusses the challenges and ethical considerations surrounding the adoption of AI in finance. Ultimately, this research underscores the critical role of AI in shaping the future of financial services, fostering resilience, and driving innovation in risk management practices.

This study explored how day-to-day social work practice with children and families in Wales responds to poverty, building on case studies already published in the other three UK nations. A case study design was used. The sites were locality teams in two local authorities, differing in their children looked after rates and trajectories of these over time. Qualitative research methods included practice observations; interviews with staff; focus groups; mapping of decision-making processes; and a sample of family case narratives. In one local authority, the range of data was similar to the other
UK nation case studies. However, in the second, data collection was adapted to the coronavirus disease-19 (COVID-19) lockdown context. Some evidence was found of narratives that emphasised the cultural, rather than material, aspect of poverty, and blamed parents for making inappropriate spending choices. Poverty alleviation was generally seen as outside of social workers’ control and requiring earlier help before social services involvement. In one of the local authorities, there was some awareness shown of the impact of poverty on parenting. In the other, that took part in the study during 2020, the exacerbating effects of families in lockdown were described, including the lack of family support due to pandemic restrictions.

The review aimed to investigate two central issues. 1.To what extent is there evidence that poverty increases the amount of child abuse and neglect (CAN), and/or affects the nature of child abuse and neglect? How does this occur, how large are these effects and to whom do they apply? 2.To what extent is there evidence that CAN increases poverty later in life, how large are these effects and to whom do they apply? Within these two issues evidence about equality and diversity, and cost were considered throughout.

The 2012 White Paper for Vulnerable Children acknowledges that children “who are at greatly elevated risk of maltreatment ... require more intensive and cross-cutting interventions to address the depth and breadth of vulnerabilities they present” (p.58). In order to intervene early to mitigate the risk of maltreatment, the White Paper includes a proposal to use predictive risk modelling (PRM) tools to assist professionals in identifying children who are at risk of abuse or neglect. The goal of using PRM tools is to support an early preventive intervention strategy. The implementation of PRM tools is subject to the outcomes of a feasibility study, ethical evaluation and trialling.

This article explores social care responses to children experiencing criminal exploitation and violence. The article draws on data from two children's social care departments in England applying a Contextual Safeguarding framework to extra-familial harm. Using evidence from interviews, focus groups, meeting observations and two peer assessments, the article explores factors that facilitate welfare approaches when children commit crimes and those conditions that undermine welfare approaches. The findings outline five conditions which facilitate or inhibit welfare responses including: whether legal rights promote the best interests of the child, if harm reduction prioritises a child's needs, if language is underpinned by caring intention, the extent that systems harm is recognised and addressed and how practitioners gain knowledge of young people. The discussion introduces a welfare framework for social workers to define and describe what the conditions for welfare responses can constitute.

If patients at risk of admission or readmission to hospital or other forms of care could be identified and offered suitable early interventions then their lives and longterm health may be improved by reducing the chances of future admission or readmission to care, and hopefully, their cost of care reduced. Considerable work has been carried out in this subject area especially in the USA and the UK. This has led for instance to the development of tools such as PARR, PARR-30, and the Combined Predictive Model for prediction of emergency readmission or admission to acute care. Here we perform a structured review the academic and grey literature on predictive risk tools for social care utilisation, as well as admission and readmission to general hospitals and psychiatric hospitals. This is the first phase of a project in partnership with Docobo Ltd and funded by Innovate UK, in which we seek to develop novel predictive risk tools and dashboards to assist commissioners in Clinical Commissioning Groups with the triangulation of the intelligence available from routinely collected data to optimise integrated care and better understand the complex needs of individuals.

The objective of this study was to develop, test and benchmark a framework and a predictive risk model for hospital emergency readmission within 12 months. We performed the development using routinely collected Hospital Episode Statistics data covering inpatient hospital admissions in England. Three different timeframes were used for training, testing and benchmarking: 1999 to 2004, 2000 to 2005 and 2004 to 2009 financial years. Each timeframe includes 20% of all inpatients admitted within the trigger year. The comparisons were made using positive predictive value, sensitivity and specificity for different risk cutoffs , risk bands and top risk segments, together with the receiver operating characteristic curve. The constructed Bayes Point Machine using this feature selection framework produces a risk probability for each admitted patient, and it was validated for different timeframes, sub-populations and cutoff points. At risk cutoff of 50%, the positive predictive value was 69.3% to 73.7%, the specificity was 88.0% to 88.9% and sensitivity was 44.5% to 46.3% across different timeframes. Also, the area under the receiver operating characteristic curve was 73.0% to 74.3%. The developed framework and model performed considerably better than existing modelling approaches with high precision and moderate sensitivity.

If patients at risk of admission or readmission to hospital or other forms of care could be identified and offered suitable early interventions then their lives and longterm health may be improved by reducing the chances of future admission or readmission to care, and hopefully, their cost of care reduced. Considerable work has been carried out in this subject area especially in the USA and the UK. This has led for instance to the development of tools such as PARR, PARR-30, and the Combined Predictive Model for prediction of emergency readmission or admission to acute care. Here we perform a structured review the academic and grey literature on predictive risk tools for social care utilisation, as well as admission and readmission to general hospitals and psychiatric hospitals. This is the first phase of a project in partnership with Docobo Ltd and funded by Innovate UK, in which we seek to develop novel predictive risk tools and dashboards to assist commissioners in Clinical Commissioning Groups with the triangulation of the intelligence available from routinely collected data to optimise integrated care and better understand the complex needs of individuals.

The objective of this study was to develop, test and benchmark a framework and a predictive risk model for hospital emergency readmission within 12 months. We performed the development using routinely collected Hospital Episode Statistics data covering inpatient hospital admissions in England. Three different timeframes were used for training, testing and benchmarking: 1999 to 2004, 2000 to 2005 and 2004 to 2009 financial years. Each timeframe includes 20% of all inpatients admitted within the trigger year. The comparisons were made using positive predictive value, sensitivity and specificity for different risk cutoffs , risk bands and top risk segments, together with the receiver operating characteristic curve. The constructed Bayes Point Machine using this feature selection framework produces a risk probability for each admitted patient, and it was validated for different timeframes, sub-populations and cutoff points. At risk cutoff of 50%, the positive predictive value was 69.3% to 73.7%, the specificity was 88.0% to 88.9% and sensitivity was 44.5% to 46.3% across different timeframes. Also, the area under the receiver operating characteristic curve was 73.0% to 74.3%. The developed framework and model performed considerably better than existing modelling approaches with high precision and moderate sensitivity.

Develop predictive models for a paediatric population that provide information for paediatricians and health authorities to identify children at risk of hospitalisation for conditions that may be impacted through improved patient care. Retrospective healthcare utilisation analysis with multivariable logistic regression models. Demographic information linked with utilisation of health services in the years 2006-2014 was used to predict risk of hospitalisation or death in 2015 using a longitudinal administrative database of 527 458 children aged 1-13 years residing in the Regione Emilia-Romagna (RER), Italy, in 2014. Models designed to predict risk of hospitalisation or death in 2015 for problems that are potentially avoidable were developed and evaluated using the C-statistic, for calibration to assess performance across levels of predicted risk, and in terms of their sensitivity, specificity and positive predictive value. Of the 527 458 children residing in RER in 2014, 6391 childre...

If patients at risk of admission or readmission to hospital or other forms of care could be identified and offered suitable early interventions then their lives and longterm health may be improved by reducing the chances of future admission or readmission to care, and hopefully, their cost of care reduced. Considerable work has been carried out in this subject area especially in the USA and the UK. This has led for instance to the development of tools such as PARR, PARR-30, and the Combined Predictive Model for prediction of emergency readmission or admission to acute care. Here we perform a structured review the academic and grey literature on predictive risk tools for social care utilisation, as well as admission and readmission to general hospitals and psychiatric hospitals. This is the first phase of a project in partnership with Docobo Ltd and funded by Innovate UK, in which we seek to develop novel predictive risk tools and dashboards to assist commissioners in Clinical Commissioning Groups with the triangulation of the intelligence available from routinely collected data to optimise integrated care and better understand the complex needs of individuals.

This article explores social care responses to children experiencing criminal exploitation and violence. The article draws on data from two children’s social care departments in England applying a Contextual Safeguarding framework to extra-familial harm. Using evidence from interviews, focus groups, meeting observations and two peer assessments, the article explores factors that facilitate welfare approaches when children commit crimes and those conditions that undermine welfare approaches. The findings outline five conditions which facilitate or inhibit welfare responses including: whether legal rights promote the best interests of the child, if harm reduction prioritises a child’s needs, if language is underpinned by caring intention, the extent that systems harm is recognised and addressed and how practitioners gain knowledge of young people. The discussion introduces a welfare framework for social workers to define and describe what the conditions for welfare responses can const...

Early intervention, promoted as being important to the prevention of child maltreatment, is challenged by the difficulty of identifying at risk families before patterns of abuse are established. A way of identifying these families before they reach the radar of statutory systems of child protection is through predictive risk modeling (PRM). Using large datasets PRM tools are able to use algorithms with significant capacity to ascertain and stratify children's risk of experiencing maltreatment in the future. In the process, however, they also identify families who may well benefit from support but are not on a maltreatment trajectorythe so called 'false positives' who would not be among those families later identified as mistreating their children. Whilst early identification of families through the use of PRM has the potential to offer opportunities to provide supportive services that could ameliorate future harm to children, it is clear that it also has the potential to mistakenly target and label families as potential child abusers. This article discusses challenges and opportunities associated with the use of PRM in child protection. It briefly discusses the development of PRM in New Zealand, and traverses some of the complex issues as systems attempt to better target limited resources in the context of fiscal restraint.

Intervention in child neglect faces many challenges to effectiveness, including: (a) The lack of a cohesive, agreed-upon conceptual/theoretical framework and inconsistent definition of the problem; (b) disjointed intervention in various components of the problem; and (c) the social justice issues of inequity based on class, race, and gender. These conditions have led to a situation in which the role of poverty, potentially one of the most important contextual factors in neglect, can be overlooked. This article presents an argument for the need to pay more explicit attention to the definition of neglect, with particular focus on the role of poverty, in order to improve intervention in this pervasive problem. It concludes with recommendations for research, policy, and practice in child welfare.

The 2012 White Paper for Vulnerable Children acknowledges that children “who are at greatly elevated risk of maltreatment ... require more intensive and cross-cutting interventions to address the depth and breadth of vulnerabilities they present” (p.58). In order to intervene early to mitigate the risk of maltreatment, the White Paper includes a proposal to use predictive risk modelling (PRM) tools to assist professionals in identifying children who are at risk of abuse or neglect. The goal of using PRM tools is to support an early preventive intervention strategy. The implementation of PRM tools is subject to the outcomes of a feasibility study, ethical evaluation and trialling.

Signs of Safety (SoS) is a comprehensive assessment framework (AF) used in some form in most English children’s service departments as well as in many other countries. The Framework for the Assessment of Children in Need and their Families (AF) was introduced nearly twenty years ago to address identified failures to adopt a holistic approach to assessments of families where there were concerns that children were in need of protection or support. The AF is not only a statutory requirement in England but it has influenced the development of approaches to assessment globally. An evaluation of SoS in pilot local authorities in England provided opportunities to examine the extent to which the two frameworks were being used together. Findings from interviews, case record analysis and a survey indicate that while recording processes were aligned with the AF, it was not central to many of the assessments conducted and it proved difficult to detect its influence on their content. It appears ...

Contact with child protection systems are a key site of the expression of social inequalities, yet research into the size and nature of this relationship remains sparse in the Aotearoa New Zealand system context. This article reports on a study of the relationships between child protection system contact and small area-level deprivation. Using a national linked dataset including all children with system contact in 2013-14, (n = 13,851 children) it found there is a marked relationship between deprivation and system contact, and significant differences between regions for all three outcomes of interest. Compared to children living in the least deprived quintile of small areas, children in the most deprived quintile had, on average, 13 times the rate of substantiation, 18 times the rate of a family group conference, and 6 times their chance of placement in foster care. There was limited evidence for the 'inverse intervention law' that proposes that children in similarly deprived small areas have higher rates of child protection system contact if they live in less deprived regions (larger areas). The pattern of placements showed the strongest support for this law, with children in similarly deprived small areas having, on average, almost twice the rate of placement if they lived in less deprived regions compared to more deprived regions. These findings have implications for policy, as they suggest a need to apply an inequalities perspective to child protection similarly to health inequities. Specifically, action is needed to address the causes of deprivation, provide services that respond to families living in poverty, address biases, and undertake further research to examine the interactions between demand and supply of services between similar deprivation levels. 1. Inequalities in child protection Many children across the globe have contact with child protection systems, that is, the institutions established in nation states to respond to child abuse and neglect. As this population grows, it has become clear that the population in contact with child protection systems is not a representative one, but is socially stratified (Bywaters, 2015. (The relationship between poverty and contact with the child protection system can be viewed through an inequalities framework. As in health where emphasis is placed on reducing disparities, viewing the relationship between deprivation and contact with the child protection system through an inequalities framework presents opportunities for improvement 'upstream' to respond to the social determinants of system contact. An inequalities framework provides a way of conceptualising and theorising the intersecting relationships between deprivation, ethnicity, location and system contact (as well as other related variables such as age and disability) (Featherstone et al., 2017; Featherstone, Gupta, Morris, & Warner, 2016). This article reports on a study examining the relationship between deprivation and child protection system contact in Aotearoa NZ, as a first step towards understanding how inequalities operate in child protection in this national environment (Keddell and Davie, 2018a). 1.1. Inequalities as a conceptual framework Variations in patterns of children's contact with child protection systems are often thought of as only reflecting institutional or family factors, but social determinants also contribute to patterns of system contact. Bywaters (2015) notes that "For years, authors have reported that deprivation is the largest factor explaining major differences between local authorities in key aspects of child protection …However, unlike the longstanding, extensive analyses of the social determinants of inequalities in health... the exploration of inequality as a key concept in child protection …has scarcely developed" ((p. 7).

The objective of this study was to develop, test and benchmark a framework and a predictive risk model for hospital emergency readmission within 12 months. We performed the development using routinely collected Hospital Episode Statistics data covering inpatient hospital admissions in England. Three different timeframes were used for training, testing and benchmarking: 1999 to 2004, 2000 to 2005 and 2004 to 2009 financial years. Each timeframe includes 20% of all inpatients admitted within the trigger year. The comparisons were made using positive predictive value, sensitivity and specificity for different risk cutoffs , risk bands and top risk segments, together with the receiver operating characteristic curve. The constructed Bayes Point Machine using this feature selection framework produces a risk probability for each admitted patient, and it was validated for different timeframes, sub-populations and cutoff points. At risk cutoff of 50%, the positive predictive value was 69.3% to 73.7%, the specificity was 88.0% to 88.9% and sensitivity was 44.5% to 46.3% across different timeframes. Also, the area under the receiver operating characteristic curve was 73.0% to 74.3%. The developed framework and model performed considerably better than existing modelling approaches with high precision and moderate sensitivity.

If patients at risk of admission or readmission to hospital or other forms of care could be identified and offered suitable early interventions then their lives and longterm health may be improved by reducing the chances of future admission or readmission to care, and hopefully, their cost of care reduced. Considerable work has been carried out in this subject area especially in the USA and the UK. This has led for instance to the development of tools such as PARR, PARR-30, and the Combined Predictive Model for prediction of emergency readmission or admission to acute care. Here we perform a structured review the academic and grey literature on predictive risk tools for social care utilisation, as well as admission and readmission to general hospitals and psychiatric hospitals. This is the first phase of a project in partnership with Docobo Ltd and funded by Innovate UK, in which we seek to develop novel predictive risk tools and dashboards to assist commissioners in Clinical Commissioning Groups with the triangulation of the intelligence available from routinely collected data to optimise integrated care and better understand the complex needs of individuals.

In the last few years, predictive risk modelling has been suggested for use in the child welfare environment as an efficient means of targeting preventive resources and improving practitioner decision-making. First raised in the green paper on vulnerable children, then translated into the white paper on vulnerable children and the Children’s Action Plan, and now part of the Child, Youth and Family review remit, this particular tool has provoked a barrage of opinions and wide-ranging analyses, concerning ethical implications, feasibility and data issues, possible uses and political consequences (Ministry of Social Development, 2011, 2012). This has resulted in a flurry of media, academic and policy debates, both here and internationally, and many reviews and related publications.

There is a growing movement to integrate conceptual tools from the health inequalities field into research that examines the relationship between inequalities and chances of child protection system contact. This article outlines the key concepts of an inequalities perspective, and discusses how these apply to inequalities in child protection in the Aotearoa New Zealand context. Drawing on existing research, this article shows that while there is evidence of links between deprivation, ethnicity, location and system contact, a more systematic research agenda shaped by an inequalities perspective would contribute to understanding more fully the social determinants of contact with the child protection system. An inequalities perspective provides balance to the current 'social investment' policy approach that targets individuals and families for service provision, with little attention to how structural inequalities impact on system contact. Directions for research are discussed, with some specific questions suggested. These include questions relating to the relationships between social inequalities and various decision points in the child protection system; if a social gradient exists and how steep it is; the interrelationship between ethnicity, deprivation and patterns of system contact; and how similarly deprived children in different locations compare with each other in relation to child protection system contact, that is, is there an 'inverse intervention law' operating?

The White Paper on Vulnerable Children before the Aotearoa/New Zealand parliament proposes changes that will significantly reconstruct the child welfare systems in this country, including the use of a predictive risk model (PRM). This article explores the ethics of this strategy in a child welfare context. Tensions exist, including significant ethical problems such as use of information without consent, breaches of privacy and stigmatisation, without clear evidence of the benefits outweighing these costs. Broader implicit assumptions about the causes of child abuse and risk and their intersections with wider discursive, political and systems design contexts are discussed. Drawing on Houston et al. (2010) this paper highlights the potential for a PRM to contribute to a neo-liberal agenda that individualises social problems, reifies risk and abuse, and narrowly prescribes service provision. However, with reference to child welfare and child protection orientations, the paper suggests ...

This article examines recent data on the costs of poverty, child abuse and neglect (CAN), paying particular attention to contemporary patterns of spending in England. The discussion explores the ‘reactive’ spending necessary to deal with the short/medium term consequences of poverty and CAN and the ‘preventative’ spending associated with some interpretations of early intervention The evidence outlined suggests that the financial costs of poverty and CAN are both substantial and associated.

This chapter critically examines the utility of a public health approach in addressing, and preventing complex constellations of maltreatment, particularly when this occurs within impoverished neighbourhoods and communities. It discusses issues surrounding the tensions between the increasing push for more accurate risk assessment against concerns relating to greater social surveillance. Both approaches relate to the value of advanced data-integration systems. It begins by examining problems in the investigation-oriented child protection system that fail to engage families, provide needed services, and proactive early intervention and prevention. A public health model seeks to address prevention issues and takes seriously the social determinants of inequality and poor outcomes for low-income families. The question of whether its whole-of-population epidemiological focus translates well to child protection is one this chapter seeks to examine.

In debates about knowledge in social work, the terms 'objectivity' and 'subjectivity' are frequently used with varying degrees of positive and negative connotations. We argue that the terms have become so ambiguous that they should be avoided. In its place, we suggest focusing on the individual attributes associated with objectivity and subjectivity and consider how the desirable attributes can be strengthened and the undesirable ones avoided. This division differs significantly from the typical objective/ subjective division. We examine three key social work issues: the contribution of empirical research, dealing with dissent and the role of the personal. When the attributes of objectivity and subjectivity are examined in detail, it becomes apparent that they vary in how desirable and how feasible they are. A more precise use of language makes it easier to see the contributions of values, bias and power in social work policy and practice and reduce the risks of people over-claiming the reliability and neutrality of their assertions.

The White Paper on Vulnerable Children before the Aotearoa/New Zealand parliament proposes changes that will significantly reconstruct the child welfare systems in this country, including the use of a predictive risk model (PRM). This article explores the ethics of this strategy in a child welfare context. Tensions exist, including significant ethical problems such as use of information without consent, breaches of privacy and stigmatisation, without clear evidence of the benefits outweighing these costs. Broader implicit assumptions about the causes of child abuse and risk and their intersections with wider discursive, political and systems design contexts are discussed. Drawing on Houston et al. (2010) this paper highlights the potential for a PRM to contribute to a neo-liberal agenda that individualises social problems, reifies risk and abuse, and narrowly prescribes service provision. However, with reference to child welfare and child protection orientations, the paper suggests ...

provides an insightful analysis of the emergence and persistence of child poverty in New Zealand (Boston, 2014, Educational Philosophy and Theory). His remarks on why child poverty matters are brief but, as he reports, "[t]here is a large and robust body of research on the harmful consequences of child poverty" (Boston 2014, pp. 10-11). One cost he does not explicitly mention is the increased risk of maltreatment faced by children living in poverty. Given the clear correlation between risk of abuse and poverty, Boston's recommendations might be expected to go some way to addressing New Zealand's appalling child maltreatment statistics. However, Boston himself identifies both fiscal and political barriers to the implementation of his proposed strategy. "Fundamentally", he observes, "without adequate multi-party agreement, it will be hard to reduce child poverty on a durable basis" and "[c]urrently, such agreement is lacking" (Boston, 2014, p. 21). Even if we accept Boston's strategies for child-poverty reduction, then, it is prudent to consider other responses to the consequences of child poverty. Significant and controversial aspects of New Zealand's 2013 White Paper for Vulnerable Children can be read in this light, and one of them-the proposal to use predictive risk modeling (PRM) to identify children at risk of maltreatment-is the focus of this commentary.

Algorithmic tools are increasingly used in child protection decision-making. Fairness considerations of algorithmic tools usually focus on statistical fairness, but there are broader justice implications relating to the data used to construct source databases, and how algorithms are incorporated into complex sociotechnical decision-making contexts. This article explores how data that inform child protection algorithms are produced and relates this production to both traditional notions of statistical fairness and broader justice concepts. Predictive tools have a number of challenging problems in the child protection context, as the data that predictive tools draw on do not represent child abuse incidence across the population and child abuse itself is difficult to define, making key decisions that become data variable and subjective. Algorithms using these data have distorted feedback loops and can contain inequalities and biases. The challenge to justice concepts is that individual...

In the last few years, predictive risk modelling has been suggested for use in the child welfare environment as an efficient means of targeting preventive resources and improving practitioner decision-making. First raised in the green paper on vulnerable children, then translated into the white paper on vulnerable children and the Children’s Action Plan, and now part of the Child, Youth and Family review remit, this particular tool has provoked a barrage of opinions and wide-ranging analyses, concerning ethical implications, feasibility and data issues, possible uses and political consequences (Ministry of Social Development, 2011, 2012). This has resulted in a flurry of media, academic and policy debates, both here and internationally, and many reviews and related publications.

The objective of this study was to develop, test and benchmark a framework and a predictive risk model for hospital emergency readmission within 12 months. We performed the development using routinely collected Hospital Episode Statistics data covering inpatient hospital admissions in England. Three different timeframes were used for training, testing and benchmarking: 1999 to 2004, 2000 to 2005 and 2004 to 2009 financial years. Each timeframe includes 20% of all inpatients admitted within the trigger year. The comparisons were made using positive predictive value, sensitivity and specificity for different risk cutoffs , risk bands and top risk segments, together with the receiver operating characteristic curve. The constructed Bayes Point Machine using this feature selection framework produces a risk probability for each admitted patient, and it was validated for different timeframes, sub-populations and cutoff points. At risk cutoff of 50%, the positive predictive value was 69.3% to 73.7%, the specificity was 88.0% to 88.9% and sensitivity was 44.5% to 46.3% across different timeframes. Also, the area under the receiver operating characteristic curve was 73.0% to 74.3%. The developed framework and model performed considerably better than existing modelling approaches with high precision and moderate sensitivity.

If patients at risk of admission or readmission to hospital or other forms of care could be identified and offered suitable early interventions then their lives and longterm health may be improved by reducing the chances of future admission or readmission to care, and hopefully, their cost of care reduced. Considerable work has been carried out in this subject area especially in the USA and the UK. This has led for instance to the development of tools such as PARR, PARR-30, and the Combined Predictive Model for prediction of emergency readmission or admission to acute care. Here we perform a structured review the academic and grey literature on predictive risk tools for social care utilisation, as well as admission and readmission to general hospitals and psychiatric hospitals. This is the first phase of a project in partnership with Docobo Ltd and funded by Innovate UK, in which we seek to develop novel predictive risk tools and dashboards to assist commissioners in Clinical Commissioning Groups with the triangulation of the intelligence available from routinely collected data to optimise integrated care and better understand the complex needs of individuals.

This article examines the viability of the Risk Predictor Model (RPM) and its counterpart the actuarial risk assessment (ARA) tool in the form of the Tuituia Assessment Framework to address child vulnerability in New Zealand. In doing so, it suggests that these types of risk-assessment tools fail to address issues of contingency and complexity at the heart of the relationship-based nature of social work practice. Such developments have considerable implications for the capacity to enhance critical reflexive practice skills, whilst the introduction of these risk tools is occurring at a time when the reflexive space is being eroded as a result of the increased regulation of practice and supervision. It is further asserted that the primary aim of such instruments is not so much to detect risk, but rather to foster professional conformity with these managerialist risk-management systems so prevalent in contemporary Western societies.

In England, the involvement of the state in private family life has significantly changed since 1989, and is about to undergo yet further transformation with the adoption of predictive risk modelling. Its implementation by at least eight local authorities and promotion by the central government operates an ontological and methodological revolution in child protection policies and practice. This article particularly discusses the conceptual and empirical limits to their use. It makes the case that the data fed to risk models, and the institutional structures through which they are run, limit the understanding of maltreatment. Based on misleading evidence, risk models are set to mistakenly equate socioeconomic disadvantage with risks, thus threatening to automatise the discrimination and alienation of the poorest sections of the population.

This article is part Eight of a series exploring the links between evaluation and technologies. It considers the rehabilitation of RCTs as a methodology through a new relationship with big data: one that repurposes the technology of RCTs to be useful within the lifecycle of the intervention. Reconsidering the application of impact evaluation provides an opportunity to apply the uncanny valley to evaluation contexts to give us a “good enough” view of results.

This article is part seven of a series exploring the links between evaluation and technologies. It examines some of the lesser and significant changes in technology applications that are on the horizon and have already happened, and what some of the major development actors are putting their coordinated efforts into. From the view of an evaluator, what does it mean to adopt and adapt? And placing it all in perspective.

This article is part six of a series exploring the links between evaluation and technologies. It examines a hypothetical intervention that was designed but not implemented to identify children at risk of maltreatment. The report of the results of the machine learning platform was followed up by insightful inquiries into the flaws, risks and ethical considerations raised by the experiment.

This article is part five of a series exploring the links between evaluation and technologies. It considers the role of ethics in the causes and effects of algorithms and whether it is in the algorithm or the data that we find bias. The human need to imbue machines with values and morality seems to create a paradox for some given our frequent nuanced minor transgressions.

This article is part four of a series exploring the links between evaluation and technologies. It considers how we deal with quantitative data, and asks whether we can make better use of data for decision making. The tensions for evaluators arise with the introduction of algorithms, and mysterious “black boxes”. But we also need to think about how humans’ assumptions and biases leak into our technologies, whether digital or analogue.

This article is part three of a series exploring the links between evaluation and technologies. It explores the gulfs between digital data-centric and analogue data-centric worlds, seen in not only the workflows but in the very fabric of the buildings where we work. In the evaluation world we collect and analyse data, but arguably it did not occur to us that the massive body of work we have produced is its own dataset, one that repels new technologies’ ability to mine it through its sheer unstructuredness. This calls for cooperation between social and data scientists to make use of it.

This article is part two of a series exploring the links between evaluation and technologies. It focusses on the significant efforts made to overcome the perennial problem of distance including through technology. Evaluators have a very specific relationship with distance in the context of objectivity. And no mini-series on technology would be complete without considering the veracity issues that the implementation of blockchains attempts to address.

This article is part one of a series exploring the links between evaluation and technologies. This installment considers the origins of cybernetics through reflections on current manifestations like the behavioural influences of algorithms, surveillance and machine learning. In thinking about how we  implement technology today, it becomes ever more necessary to revisit old and revise present philosophical and ethical conundrums.

If patients at risk of admission or readmission to hospital or other forms of care could be identified and offered suitable early interventions then their lives and long-term health may be improved by reducing the chances of future admission or readmission to care, and hopefully, their cost of care reduced.

Considerable work has been carried out in this subject area especially in the USA and the UK. This has led for instance to the development of tools such as PARR, PARR-30, and the Combined Predictive Model for prediction of emergency readmission or admission to acute care.

Here we perform a structured review the academic and grey literature on predictive risk tools for social care utilisation, as well as admission and readmission to general hospitals and psychiatric hospitals. This is the first phase of a project in partnership with Docobo Ltd and funded by Innovate UK, in which we seek to develop novel predictive risk tools and dashboards to assist commissioners in Clinical Commissioning Groups with the triangulation of the intelligence available from routinely collected data to optimise integrated care and better understand the complex needs of individuals.

The objective of this study was to develop, test and benchmark a framework and a predictive risk model for hospital emergency readmission within 12 months. We performed the development using routinely collected Hospital Episode Statistics data covering inpatient hospital admissions in England. Three different timeframes were used for training, testing and benchmarking: 1999 to 2004, 2000 to 2005 and 2004 to 2009 financial years. Each timeframe includes 20% of all inpatients admitted within the trigger year.

The comparisons were made using positive predictive value, sensitivity and specificity for different risk cut-offs, risk bands and top risk segments, together with the receiver operating characteristic curve. The constructed Bayes Point Machine using this feature selection framework produces a risk probability for each admitted patient, and it was validated for different timeframes, sub-populations and cut-off points. At risk cut-off of 50%, the positive predictive value was 69.3% to 73.7%, the specificity was 88.0% to 88.9% and sensitivity was 44.5% to 46.3% across different timeframes. Also, the area under the receiver operating characteristic curve was 73.0% to 74.3%. The developed framework and model performed considerably better than existing modelling approaches with high precision and moderate sensitivity.

Early intervention, promoted as being important to the prevention of child maltreatment, is challenged by the difficulty of identifying at risk families before patterns of abuse are established. Away of identifying these families before they reach the radar of statutory systems of child protection is through predictive risk modeling (PRM). Using large datasets PRM tools are able to use algorithms with significant capacity to ascertain and stratify children's risk of experiencing maltreatment in the future. In the process, however, they also identify families who may well benefit from support but are not on a maltreatment trajectory — the so called ‘false positives’
who would not be among those families later identified as mistreating their children.Whilst early identification of families through the use of PRM has the potential to offer opportunities to provide supportive services that could ameliorate future harm to children, it is clear that it also has the potential to mistakenly target and label families as potential child abusers. This article discusses challenges and opportunities associated with the use of PRM in child protection. It briefly discusses the development of PRM in New Zealand, and traverses some of
the complex issues as systems attempt to better target limited resources in the context of fiscal restraint.