Predicting individual cases of major adolescent psychiatric conditions with artificial intelligence

Three-quarters of lifetime mental illness occurs by the age of 24, but relatively little is known about how to robustly identify youth at risk to target intervention efforts known to improve outcomes. Barriers to knowledge have included obtaining robust predictions while simultaneously analyzing large numbers of different types of candidate predictors. In a new, large, transdiagnostic youth sample and multidomain high-dimension data, we used 160 candidate predictors encompassing neural, prenatal, developmental, physiologic, sociocultural, environmental, emotional and cognitive features and leveraged three different machine learning algorithms optimized with a novel artificial intelligence meta-learning technique to predict individual cases of anxiety, depression, attention deficit, disruptive behaviors and post-traumatic stress. Our models tested well in unseen, held-out data (AUC ≥ 0.94). By utilizing a large-scale design and advanced computational approaches, we were able to compare the relative predictive ability of neural versus psychosocial features in a principled manner and found that psychosocial features consistently outperformed neural metrics in their relative ability to deliver robust predictions of individual cases. We found that deep learning with artificial neural networks and tree-based learning with XGBoost outperformed logistic regression with ElasticNet, supporting the conceptualization of mental illnesses as multifactorial disease processes with non-linear relationships among predictors that can be robustly modeled with computational psychiatry techniques. To our knowledge, this is the first study to test the relative predictive ability of these gold-standard algorithms from different classes across multiple mental health conditions in youth within the same study design in multidomain data utilizing >100 candidate predictors. Further research is suggested to explore these findings in longitudinal data and validate results in an external dataset.

The multifaceted consequences and economic costs of child anxiety problems: A systematic review and meta-analysis

Background

Over a quarter of people have an anxiety disorder at some point in their life, with many first experiencing difficulties during childhood or adolescence. Despite this, gaps still exist in the current evidence base of the multiple consequences of childhood anxiety problems and their costs.

Methods

A systematic review of Medline, PsycINFO, EconLit and the National Health Service Economic Evaluation Database was conducted for longitudinal and economic studies reporting on the association between childhood anxiety problems and at least one individual-, family- or societal-level outcome or cost. All studies were synthesised narratively. For longitudinal studies, 'effect direction' was used as a common metric, with random effects meta-analysis undertaken where possible.

Results

Eighty-three studies met inclusion criteria and were synthesised narratively. We identified 788 separate analyses from the longitudinal studies, which we grouped into 15 overarching outcome domains. Thirteen of the studies were incorporated into 13 meta-analyses, which indicated that childhood anxiety disorders were associated with future anxiety, mood, behaviour and substance disorders. Narrative synthesis also suggested associations between anxiety problems and worse physical health, behaviour, self-harm, eating, relationship, educational, health care, employment, and financial outcomes. 'Effect direction' was conflicting in some domains due to a sparse evidence base. Higher economic costs were identified for the child, their families, healthcare providers and wider society, although evidence was limited and only covered short follow-up periods, up to a maximum of 2 years. Total annual societal costs per anxious child were up to £4040 (2021 GBP).

Conclusions

Childhood anxiety problems are associated with impaired outcomes in numerous domains, and considerable economic costs, which highlight the need for cost-effective interventions and policies to tackle them. More economic evidence is needed to inform models of the long-term, economic-related, consequences of childhood anxiety problems.

A Novel and Integrated Digitally Supported System of Care for Depression and Anxiety: Findings From an Open Trial

BACKGROUND

The global burden of anxiety and depression has created an urgent need for scalable approaches to increase access to evidence-based mental health care. The Screening and Treatment for Anxiety and Depression (STAND) system of care was developed to meet this need through the use of internet-connected devices for assessment and provision of treatment. STAND triages to level of care (monitoring only, digital therapy with coaches, digital therapy assisted by clinicians in training, and clinical care) and then continuously monitors symptoms to adapt level of care. Triaging and adaptation are based on symptom severity and suicide risk scores obtained from computerized adaptive testing administered remotely.

OBJECTIVE

This article discusses how the STAND system of care improves upon current clinical paradigms, and presents preliminary data on feasibility, acceptability, and effectiveness of STAND in a sample of US-based university students.

METHODS

US-based university students were recruited and enrolled in an open trial of the STAND system of care. Participants were triaged based on initial symptom severity derived from a computerized adaptive test and monitored over 40 weeks on anxiety, depression, and suicide risk to inform treatment adaptation and evaluate preliminary effectiveness.

RESULTS

Nearly 5000 students were screened and 516 received care. Depression and anxiety severity scores improved across all tiers (P<.001 in all cases). Suicide risk severity improved in the highest tier (ie, clinical care; P<.001). Acceptability and feasibility were demonstrated.

CONCLUSIONS

STAND is a feasible and acceptable model of care that can reach large numbers of individuals. STAND showed preliminary effectiveness on all primary outcome measures. Current directions to improve STAND are described.

Review: Economic evidence of preventive interventions for anxiety disorders in children and adolescents - a systematic review

BACKGROUND

Anxiety disorders are common in children and youth. Also, in prevention, be it universal, selective or indicated, economic evaluation supports decision-making in the allocation of scarce resources. This review identified and summarised the existing evidence of economic evaluations for the prevention of anxiety disorders in children and adolescents.

METHODS

A systematic search was conducted on the EBSCO, Scopus, Web of Science, ProQuest, Cochrane and PubMed databases. We included studies that focused on children and adolescents under 18 years of age, aimed to prevent anxiety disorders and presented an incremental analysis of costs and effectiveness. A registered checklist was used that assessed the quality of the included articles.

RESULTS

The search yielded 1697 articles. Five articles were included in this review. Three were RCT-based, and two were model-based studies. Out of five included interventions, one was a universal school-based intervention, two selective interventions and two indicated interventions. Universal school-based prevention of anxiety was not cost-effective compared with usual teaching. Selective parent training and indicative child- and parent-focused CBT prevention were likely cost-effective compared with usual care or doing nothing.

CONCLUSION

Parent education and cognitive behaviour therapy interventions can be cautiously interpreted as being a cost-effective way of preventing anxiety in children and adolescents. However, the evidence is weakly related to cost-effectiveness as there are only a few studies, with relatively small sample sizes and short follow-ups.

Cost-effectiveness of a transdiagnostic psychotherapy program for youth with common mental health problems

OBJECTIVES

Our objective was to evaluate the cost-effectiveness of the transdiagnostic psychotherapy program Mind My Mind (MMM) for youth with common mental health problems using a cost-utility analysis (CUA) framework and data from a randomized controlled trial. Furthermore, we analyzed the impact of the choice of informant for both quality-of-life reporting and preference weights on the Incremental Cost-Effectiveness Ratio (ICER).

METHODS

A total of 396 school-aged (6-16 years) youth took part in the 6-month trial carried out in Denmark. CUAs were carried out for the trial period and four one-year extrapolation scenarios. Costs were based on a combination of budget and self-reported costs. Youths and parents were asked to report on the youth's quality-of-life three times during the trial using the Child Health Utility 9D (CHU9D). Parental-reported CHU9D was used in the base case together with preference weights of a youth population. Analyses using self-reported CHU9D and preference weights of an adult population were also carried out.

RESULTS

The analysis of the trial period resulted in an ICER of €170,465. The analyses of the one-year scenarios resulted in ICERs between €23,653 and €50,480. The ICER increased by 24% and 71% compared to the base case when using self-reported CHU9D and adult preference weights, respectively.

CONCLUSION

The MMM intervention has the potential to be cost-effective, but the ICER is dependent on the duration of the treatment effects. Results varied significantly with the choice of informant and the choice of preference weights indicating that both factors should be considered when assessing CUA involving youth.

The youth online training and employment system: Study protocol for a randomized controlled trial of an online vocational intervention for young people with mental ill health

AIM

People diagnosed with mental disorders experience higher rates of unemployment than those without. Career adaptability, defined as the ability to respond flexibly and make informed career decisions in work and throughout career transitions, is becoming increasingly important as the nature of work changes rapidly. Early vocational intervention may ameliorate poor education and employment outcomes experienced by young people with mental ill-health and promote transferable skills and adaptability. Online-based career support allows for ongoing access throughout different career stages. The current study combines mental health-informed digital career and peer motivation, to create a Youth Online Training and Employment System (YOTES) that supports young people with mental ill-health obtain and remain in education or employment.

METHODS

This study is an unblinded randomized controlled trial for young people with mental ill-health, aged 16-25, who are seeking vocational support. Participants will be randomized to receive either YOTES, a moderated, online intervention with vocational, social, and peer motivation, or a control intervention, the headspace Digital Work and Study Service. Both groups will have access to in-person career support if seeking employment. The primary outcome will be career adaptability compared between the YOTES and control groups at 6-months post baseline. Secondary outcomes include number of hours worked in the past 7 days, hope, career confidence, psychological distress and health economic outcomes at 6- and 12-months post baseline.

CONCLUSION

Results will demonstrate whether an online career intervention moderated by career practitioners with peer motivation can result in improved career adaptability in young people with mental ill-health.

Forecasting the Suitability of Online Mental Health Information for Effective Self-Care Developing Machine Learning Classifiers Using Natural Language Features

Background: Online mental health information represents important resources for people living with mental health issues. Suitability of mental health information for effective self-care remains understudied, despite the increasing needs for more actionable mental health resources, especially among young people. Objective: We aimed to develop Bayesian machine learning classifiers as data-based decision aids for the assessment of the actionability of credible mental health information for people with mental health issues and diseases. Methods: We collected and classified creditable online health information on mental health issues into generic mental health (GEN) information and patient-specific (PAS) mental health information. GEN and PAS were both patient-oriented health resources developed by health authorities of mental health and public health promotion. GENs were non-classified online health information without indication of targeted readerships; PASs were developed purposefully for specific populations (young, elderly people, pregnant women, and men) as indicated by their website labels. To ensure the generalisability of our model, we chose to develop a sparse Bayesian machine learning classifier using Relevance Vector Machine (RVM). Results: Using optimisation and normalisation techniques, we developed a best-performing classifier through joint optimisation of natural language features and min-max normalisation of feature frequencies. The AUC (0.957), sensitivity (0.900), and specificity (0.953) of the best model were statistically higher (p < 0.05) than other models using parallel optimisation of structural and semantic features with or without feature normalisation. We subsequently evaluated the diagnostic utility of our model in the clinic by comparing its positive (LR+) and negative likelihood ratios (LR−) and 95% confidence intervals (95% C.I.) as we adjusted the probability thresholds with the range of 0.1 and 0.9. We found that the best pair of LR+ (18.031, 95% C.I.: 10.992, 29.577) and LR− (0.100, 95% C.I.: 0.068, 0.148) was found when the probability threshold was set to 0.45 associated with a sensitivity of 0.905 (95%: 0.867, 0.942) and specificity of 0.950 (95% C.I.: 0.925, 0.975). These statistical properties of our model suggested its applicability in the clinic. Conclusion: Our study found that PAS had significant advantage over GEN mental health information regarding information actionability, engagement, and suitability for specific populations with distinct mental health issues. GEN is more suitable for general mental health information acquisition, whereas PAS can effectively engage patients and provide more effective and needed self-care support. The Bayesian machine learning classifier developed provided automatic tools to support decision making in the clinic to identify more actionable resources, effective to support self-care among different populations.

Can We Build 'Somewhere That You Want to Go'? Conducting Collaborative Mental Health Service Design with New Zealand's Young People

Globally, young people are at high risk of mental health problems, but have poor engagement with services. Several international models have emerged seeking to address this gap by providing youth-specific care designed in collaboration with young people. In this study, 94 young people in New Zealand participated in collaborative workshops exploring their vision of an ideal mental health service. Participants were aged 16-25. Reflexive thematic analysis was used to identify seven themes. These describe the ideal mental health service for these young people as comfortable, accessible, welcoming, embedded in the community, holistic, adaptable, and youth-focused. In addition to describing how services might better serve the needs of youth, this article outlines a method for adapting international principles for youth-friendly care to the specific needs of a population of young people. This article provides supporting evidence that services should consider how to improve their engagement with youth through collaboration with local populations of young people.

Right care, first time: a highly personalised and measurement-based care model to manage youth mental health

Mood and psychotic syndromes most often emerge during adolescence and young adulthood, a period characterised by major physical and social change. Consequently, the effects of adolescent-onset mood and psychotic syndromes can have long term consequences. A key clinical challenge for youth mental health is to develop and test new systems that align with current evidence for comorbid presentations and underlying neurobiology, and are useful for predicting outcomes and guiding decisions regarding the provision of appropriate and effective care. Our highly personalised and measurement-based care model includes three core concepts: ▶ A multidimensional assessment and outcomes framework that includes: social and occupational function; self-harm, suicidal thoughts and behaviour; alcohol or other substance misuse; physical health; and illness trajectory. ▶ Clinical stage. ▶ Three common illness subtypes (psychosis, anxious depression, bipolar spectrum) based on proposed pathophysiological mechanisms (neurodevelopmental, hyperarousal, circadian). The model explicitly aims to prevent progression to more complex and severe forms of illness and is better aligned to contemporary models of the patterns of emergence of psychopathology. Inherent within this highly personalised approach is the incorporation of other evidence-based processes, including real-time measurement-based care as well as utilisation of multidisciplinary teams of health professionals. Data-driven local system modelling and personalised health information technologies provide crucial infrastructure support to these processes for better access to, and higher quality, mental health care for young people. CHAPTER 1: MULTIDIMENSIONAL OUTCOMES IN YOUTH MENTAL HEALTH CARE: WHAT MATTERS AND WHY?: Mood and psychotic syndromes present one of the most serious public health challenges that we face in the 21st century. Factors including prevalence, age of onset, and chronicity contribute to substantial burden and secondary risks such as alcohol or other substance misuse. Mood and psychotic syndromes most often emerge during adolescence and young adulthood, a period characterised by major physical and social change; thus, effects can have long term consequences. We propose five key domains which make up a multidimensional outcomes framework that aims to address the specific needs of young people presenting to health services with emerging mental illness. These include social and occupational function; self-harm, suicidal thoughts and behaviours; alcohol or other substance misuse; physical health; and illness type, stage and trajectory. Impairment and concurrent morbidity are well established in young people by the time they present for mental health care. Despite this, services and health professionals tend to focus on only one aspect of the presentation - illness type, stage and trajectory - and are often at odds with the preferences of young people and their families. There is a need to address the disconnect between mental health, physical health and social services and interventions, to ensure that youth mental health care focuses on the outcomes that matter to young people. CHAPTER 2: COMBINING CLINICAL STAGE AND PATHOPHYSIOLOGICAL MECHANISMS TO UNDERSTAND ILLNESS TRAJECTORIES IN YOUNG PEOPLE WITH EMERGING MOOD AND PSYCHOTIC SYNDROMES: Traditional diagnostic classification systems for mental disorders map poorly onto the early stages of illness experienced by young people, and purport categorical distinctions that are not readily supported by research into genetic, environmental and neurobiological risk factors. Consequently, a key clinical challenge in youth mental health is to develop and test new classification systems that align with current evidence on comorbid presentations, are consistent with current understanding of underlying neurobiology, and provide utility for predicting outcomes and guiding decisions regarding the provision of appropriate and effective care. This chapter outlines a transdiagnostic framework for classifying common adolescent-onset mood and psychotic syndromes, combining two independent but complementary dimensions: clinical staging, and three proposed pathophysiological mechanisms. Clinical staging reflects the progression of mental disorders and is in line with the concept used in general medicine, where more advanced stages are associated with a poorer prognosis and a need for more intensive interventions with a higher risk-to-benefit ratio. The three proposed pathophysiological mechanisms are neurodevelopmental abnormalities, hyperarousal and circadian dysfunction, which, over time, have illness trajectories (or pathways) to psychosis, anxious depression and bipolar spectrum disorders, respectively. The transdiagnostic framework has been evaluated in young people presenting to youth mental health clinics of the University of Sydney's Brain and Mind Centre, alongside a range of clinical and objective measures. Our research to date provides support for this framework, and we are now exploring its application to the development of more personalised models of care. CHAPTER 3: A COMPREHENSIVE ASSESSMENT FRAMEWORK FOR YOUTH MENTAL HEALTH: GUIDING HIGHLY PERSONALISED AND MEASUREMENT-BASED CARE USING MULTIDIMENSIONAL AND OBJECTIVE MEASURES: There is an urgent need for improved care for young people with mental health problems, in particular those with subthreshold mental disorders that are not sufficiently severe to meet traditional diagnostic criteria. New comprehensive assessment frameworks are needed to capture the biopsychosocial profile of a young person to drive highly personalised and measurement-based mental health care. We present a range of multidimensional measures involving five key domains: social and occupational function; self-harm, suicidal thoughts and behaviours; alcohol or other substance misuse; physical health; and illness type, stage and trajectory. Objective measures include: neuropsychological function; sleep-wake behaviours and circadian rhythms; metabolic and immune markers; and brain structure and function. The recommended multidimensional measures facilitate the development of a comprehensive clinical picture. The objective measures help to further develop informative and novel insights into underlying pathophysiological mechanisms and illness trajectories to guide personalised care plans. A panel of specific multidimensional and objective measures are recommended as standard clinical practice, while others are recommended secondarily to provide deeper insights with the aim of revealing alternative clinical paths for targeted interventions and treatments matched to the clinical stage and proposed pathophysiological mechanisms of the young person. CHAPTER 4: PERSONALISING CARE OPTIONS IN YOUTH MENTAL HEALTH: USING MULTIDIMENSIONAL ASSESSMENT, CLINICAL STAGE, PATHOPHYSIOLOGICAL MECHANISMS, AND INDIVIDUAL ILLNESS TRAJECTORIES TO GUIDE TREATMENT SELECTION: New models of mental health care for young people require that interventions be matched to illness type, clinical stage, underlying pathophysiological mechanisms and individual illness trajectories. Narrow syndrome-focused classifications often direct clinical attention away from other key factors such as functional impairment, self-harm and suicidality, alcohol or other substance misuse, and poor physical health. By contrast, we outline a treatment selection guide for early intervention for adolescent-onset mood and psychotic syndromes (ie, active treatments and indicated and more specific secondary prevention strategies). This guide is based on experiences with the Brain and Mind Centre's highly personalised and measurement-based care model to manage youth mental health. The model incorporates three complementary core concepts: ▶A multidimensional assessment and outcomes framework including: social and occupational function; self-harm, suicidal thoughts and behaviours; alcohol or other substance misuse; physical health; and illness trajectory. ▶Clinical stage. ▶Three common illness subtypes (psychosis, anxious depression, bipolar spectrum) based on three underlying pathophysiological mechanisms (neurodevelopmental, hyperarousal, circadian). These core concepts are not mutually exclusive and together may facilitate improved outcomes through a clinical stage-appropriate and transdiagnostic framework that helps guide decisions regarding the provision of appropriate and effective care options. Given its emphasis on adolescent-onset mood and psychotic syndromes, the Brain and Mind Centre's model of care also respects a fundamental developmental perspective - categorising childhood problems (eg, anxiety and neurodevelopmental difficulties) as risk factors and respecting the fact that young people are in a period of major biological and social transition. Based on these factors, a range of social, psychological and pharmacological interventions are recommended, with an emphasis on balancing the personal benefit-to-cost ratio. CHAPTER 5: A SERVICE DELIVERY MODEL TO SUPPORT HIGHLY PERSONALISED AND MEASUREMENT-BASED CARE IN YOUTH MENTAL HEALTH: Over the past decade, we have seen a growing focus on creating mental health service delivery models that better meet the unique needs of young Australians. Recent policy directives from the Australian Government recommend the adoption of stepped-care services to improve the appropriateness of care, determined by severity of need. Here, we propose that a highly personalised approach enhances stepped-care models by incorporating clinical staging and a young person's current and multidimensional needs. It explicitly aims to prevent progression to more complex and severe forms of illness and is better aligned to contemporary models of the patterns of emergence of psychopathology. Inherent within a highly personalised approach is the incorporation of other evidence-based processes, including real-time measurement-based care and use of multidisciplinary teams of health professionals. Data-driven local system modelling and personalised health information technologies provide crucial infrastructure support to these processes for better access to, and higher quality of, mental health care for young people.

Economics and mental health: the current scenario

Economics and mental health are intertwined. Apart from the accumulating evidence of the huge economic impacts of mental ill-health, and the growing recognition of the effects that economic circumstances can exert on mental health, governments and other budget-holders are putting increasing emphasis on economic data to support their decisions. Here we consider how economic evaluation (including cost-effectiveness analysis, cost-utility analysis and related techniques) can contribute evidence to inform the development of mental health policy strategies, and to identify some consequences at the treatment or care level that are of relevance to service providers and funding bodies. We provide an update and reflection on economic evidence relating to mental health using a lifespan perspective, analyzing costs and outcomes to shed light on a range of pressing issues. The past 30 years have witnessed a rapid growth in mental health economics, but major knowledge gaps remain. Across the lifespan, clearer evidence exists in the areas of perinatal depression identification-plus-treatment; risk-reduction of mental health problems in childhood and adolescence; scaling up treatment, particularly psychotherapy, for depression; community-based early intervention and employment support for psychosis; and cognitive stimulation and multicomponent carer interventions for dementia. From this discussion, we pull out the main challenges that are faced when trying to take evidence from research and translating it into policy or practice recommendations, and from there to actual implementation in terms of better treatment and care.

Global research priorities for youth mental health

AIM

Over the past two decades, the youth mental health field has expanded and advanced considerably. Yet, mental disorders continue to disproportionately affect adolescents and young adults. Their prevalence and associated morbidity and mortality in young people have not substantially reduced, with high levels of unmet need and poor access to evidence-based treatments even in high-income countries. Despite the potential return on investment, youth mental disorders receive insufficient funding. Motivated by these continual disparities, we propose a strategic agenda for youth mental health research.

METHOD

Youth mental health experts and funders convened to develop youth mental health research priorities, via thematic roundtable discussions, that address critical evidence-based gaps.

RESULTS

Twenty-one global youth mental health research priorities were developed, including population health, neuroscience, clinical staging, novel interventions, technology, socio-cultural factors, service delivery, translation and implementation.

CONCLUSIONS

These priorities will focus attention on, and provide a basis for, a systematic and collaborative strategy to globally improve youth mental health outcomes.

Targeting mental health care attributes by diagnosis and clinical stage: the views of youth mental health clinicians

OBJECTIVES

To explore the potential utility of clinical stage and mental disorder categories as a basis for determining which attributes of youth mental health care should be offered to which groups of young people.

METHODS

In June 2017, we conducted an online survey of youth mental health clinicians that collected information on the participants' background and areas of expertise, then presented vignettes describing young people with different stages of six mental disorders (disorder-based vignettes were matched to participants' area of expertise). For each vignette, participants were asked to give a quantitative estimate of the proportion of young people with similar mental health problems they thought would clinically benefit from each of twelve attributes of mental health care (other than pharmacological or individual psychological therapies). Survey results were analysed as independent, disorder-based samples, using standard statistical tests of significance, and as a stratified sample using mixed-effects models.

RESULTS

A total of 412 clinicians working in 32 countries participated in both parts of the survey. Respondents represented a broad range of clinical disciplines, settings and areas of expertise. Their estimated proportions of young people who would benefit from the mental health care attributes varied by clinical stage and disorder (eg, a mean of 93% [interquartile range (IQR), 90%-100%] of young people with Stage 2 psychosis were estimated to benefit from case management with a multidisciplinary team; while only 15% [IQR, 1%-25%] of young people with Stage 1b generalised anxiety disorder were estimated to benefit from collection and processing of biological samples). Neither the background of the respondents nor the sex of the characters in the vignettes significantly influenced the results.

CONCLUSION

A combination of clinical stage and disorder information might be an appropriate basis for ensuring that the right attributes of early intervention mental health care are provided to the right young people at the right time. Policy and research priorities include trialling novel services, preferences research among young people, strengthening service responses to subthreshold disorders and promoting high-fidelity collection of clinical stage data in youth mental health settings.