USING MACHINE LEARNING TO PREDICT OUTCOMES FROM ICBT
Feedback-informed therapy (FIT), where the therapist is given feedback on the progress of the client, has been shown to enhance traditional CBT, as well as internet-delivered CBT. In recent years, FIT has been benefitting from advances in machine learning (ML), which allows for more complex models to be built to assist in the detection, diagnosis, and treatment of mental health problems. ML techniques have been used to predict outcomes from treatment, select the best treatment for a given patient, identify different subgroups or subtypes of treatment response, and predict new onsets of mental health episodes. Using ML to enhance FIT initiatives, especially for technology-delivered interventions such as iCBT and within service settings that routinely collect outcome data could prove beneficial to overall service performance and client outcomes. We have recently developed one such algorithm that uses data from the first three assessments to predict whether a patient is likely to improve or not at the end of treatment. The current work will utilise a randomized controlled trial design, where the supporters (Psychological Wellbeing Practitioners) at one IAPT site (Berkshire Talking Therapies) will be randomized into two groups; supporters in group one will benefit from the predictive outcomes from the Deep Learning Model DLM while supporters in group 2 will practice as usual. The aim is to assess the performance and acceptability of this prediction tool in iCBT for depression and anxiety. Regarding performance, we expect a higher percentage of patients whose PWPs are in the DLM group to achieve reliable improvement compared to the patients whose PWPs do not have access to the tool. Regarding acceptance, we expect PWPs to find the tool acceptable, user engagement metrics to be higher in the DLM group, and the tool to influence the clinical decision-making and therapeutic skills of the PWPs.
Trial Registration: ISRCTN18059067
COMPARING LOW-INTENSITY PSYCHOLOGICAL INTERVENTIONS
Low-intensity interventions for common mental disorders (CMD) address issues such as clinician shortages and barriers to accessing care. However, there is a lack of research into their comparative effectiveness in routine care. We aimed to compare the treatment effects of three such interventions, utilising four years’ worth of routine clinical data. Users completing a course of guided self-help bibliotherapy (GSH), internet-delivered cognitive behavioural therapy (iCBT), or psychoeducational group therapy (PGT) from a stepped-care service within the NHS in England were included. Propensity score models (stratification and weighting) were used to control for allocation bias and determine the average treatment effect (ATE) between the interventions. 21,215 users comprised the study sample (GSH = 12,896, iCBT = 6,862, PGT = 1,457). Adherence-to-treatment rates were higher in iCBT. All interventions showed significant improvements in depression (PHQ-9), anxiety (GAD-7) and functioning (WSAS) scores, with the largest effect sizes for iCBT. Both propensity score models showed a significant ATE in favour of iCBT vs GSH and PGT, and in favour of GSH vs PGT. Discernible differences in effectiveness were seen for iCBT in comparison to GSH and PGT. Given the variance in delivery mode and human resources between different low-intensity interventions, findings should be considered for service provision and policy decision-making.
Research has shown that internet-based Cognitive Behavioural Therapy(iCBT) can be a very promising solution to increase access to and the dissemination of evidence-based treatments to all the populations in need. However, iCBT is still underutilised in clinical contexts, such as primary care. To achieve the effective implementation of these protocols more studies in ecological settings are needed. The Unified Protocol (UP) is a transdiagnostic CBT protocol for the treatment of emotional disorders, which includes depression, anxiety, and related disorders, that has shown its efficacy across different contexts and populations. An internet-based UP (iUP) programme has recently been developed as an emerging internet-based treatment for emotional disorders. However, the internet-delivered version of the UP (iUP) has not yet been examined empirically. using a parallel-group randomised design, the current project seeks to analyse the effectiveness of the iUP as a treatment for depression, anxiety, and related emotional disorders in a primary care public health setting.
Trial registration: ISRCTN18056450
PRECISION IN PSYCHIATRY (PIPS)
Whilst the evidence that internet-delivered cognitive behavioural therapy (iCBT) is effective in treating depression andanxiety is well-established, not all patients benefit from the treatment. There is a lack of understanding of what constitutes the optimal patient profile for iCBT. Research to-date has yielded inconsistent findings, which may partially be due to the small samples typically employed in basic analyses that are ill-equipped to interpret the complexinterplay between psychiatric risk factors and treatment. This project seeks to remedy this by leveraging large patient samples to develop objective, data-driven tools that can improve precision in iCBT treatment allocation, by generating machine learning algorithms to predict the likelihood of an individual patient’s response to iCBT. Predictors will include socio-demographics, clinical characteristics, and cognitive test data linked to specific brain mechanisms relating to psychiatric disease and treatment action. Machine learning will be used to develop models using baseline data (before treatment starts) or early clinical change data (the first 4 weeks of iCBT). Complementing this predictive approach with network analysis aims at understanding why iCBT works for some, but not others; to examine the dynamic interactions between symptoms over time and how that varies depending on treatment success. The findings will provide a clinically-valuable set of tools that can augment clinicians’ decision-making in treatment referral and improve our understanding of why iCBT works, thus informing future treatment development.
Lee, C.T., Palacios, J., Richards, D., Hanlon, A.K., Lynch, K., Harty, S., Claus, N., O’Keane, V., Stephen, K.O., Gillan, C., (2022, March 3). The Precision in Psychiatry (PIP) study: an internet-based methodology for accelerating research in treatment prediction and personalisation. https://doi.org/10.31234/osf.io/4j6z8
Fox, C., Lee, C. T., Seow, T. X. F., Lynch, K., Harty, S., Richards, D., Palacios, J., O’Keane, V., Stephan, K. E., & Gillan, C. M. (2021). Are Metacognitive Biases in Anxious-Depression Ameliorated Following Successful Depression Treatment? A Longitudinal, Observational Study. Biological Psychiatry, 89(9, Supplement), S127.
SMARTWATCH AND DEPRESSION TREATMENT
Mood tracking is commonly employed within a range of mental health interventions. Physical activity and sleep are also important for contextualizing mood data but can be difficult to track manually and rely on retrospective recall. Smartwatches could enhance self-monitoring by addressing difficulties in recall of sleep and physical activity and reducing the burden on patients in terms of remembering to track and the effort of tracking. The study has the potential to show that smartwatches are an acceptable means for patient self-monitoring within iCBT interventions for depression and support potential use-cases for smartwatches in the context of mental health interventions in general.
Nadal, C., Earley, C., Enrique, A, Vigano, N., Sas, C., Richards, D., Doherty, G. (2021). Integration of a smartwatch within an internet-delivered intervention for depression: Protocol for a feasibility randomized controlled trial on acceptance. Contemporary Clinical Trials. 103:106323. doi: 10.1016/j.cct.2021.106323. PMID: 33621632. Link to Paper
DEBT AND MENTAL HEALTH
Previous research has shown a strong relationship between financial difficulties and mental health problems. Psychological factors such as hope and worry about finances appear to be an important factor in this relationship. We developed an online based psychological intervention (Space from Money Worries) to tackle the psychological mechanisms underlying the relationship between poor mental health and financial difficulties, and to conduct an initial evaluation of the acceptability and preliminary efficacy of the intervention.
Richardson, T., Enrique, A., Earley, C., Adegoke, A., Hiscock, D., Richards, D., (2022). The Acceptability and Initial Effectiveness of ‘Space from Money Worries’: An Online Cognitive Behavioural Therapy Intervention to Tackle the Link Between Financial Difficulties and Poor Mental Health. Frontiers in Public Health. 10. Link to Paper
CANCER CARE AND ICBT
Depression and anxiety are common problems among breast cancer survivors. Carer support is one of the most important determinants of women’s psychological wellbeing. Survivors’ distress can be alleviated by giving carers access to survivors’ evidence-based treatment, which will help carers understand what survivors have been going through and help survivors feel more supported. Given the limited access to evidence-based treatments, an adapted internet-delivered cognitive behavioural therapy (iCBT) intervention for breast cancer survivors, but also open for carers’ access, has the potential to decrease survivors’ depression and anxiety symptoms and improve cancer-related communication and relationship quality between survivors and carers.
Akkol-Solakoglu, S., Hevey, D., Richards, D. (2021). A randomised controlled trial of an adapted internet-delivered cognitive behavioural therapy (iCBT) with main carer access for depression and anxiety among breast cancer survivors: study protocol. Internet Interventions. 24, 100367. Link to Paper
Development and evaluation of internet-supported treatment for recently arrived children and adolescents with psychiatric problems
The overarching aim of this research is to develop and test methods for effective screening and assessment of mental and somatic health among recently arrived immigrant children and adolescents as well as develop effective internet-delivered cognitive behavioural therapy (ICBT) interventions targeted towards common mental health problems in these groups such as anxiety, depression and post-traumatic stress. The research project is a collaboration between groups in Linköping, Stockholm, Östersund, and Dublin and will be organized in seven different work packages (WP) over a period of six years. These WPs will include development and pilot testing of three different ICBT programs, two targeted towards adolescents speaking Arabic and Dari, and the third targeted towards Arabic-speaking children aged 8-12 years and their parents. The programs targeted towards adolescents will be evaluated quantitatively first in a pilot study and secondly in a state-of-the-art randomized controlled trial whereas the program targeted towards children and their parents will be evaluated in a pilot study only. We will also conduct qualitative interviews with participants from these programs to complement the quantitative evaluation. This project aims to further the understanding regarding how to effectively assess and treat mental health problems in this vulnerable population, thus giving these young people increased well-being and a better chance of successful integration into society. https://www.swecris.se/betasearch/details/project/201805827VR?lang=en
INSOMNIA/ SLEEP DISORDER
Patients (n = 35) will be invited to use the supported intervention over an 8-week period. Participants will indicate consent and complete screening measures online before beginning the programme. They will then be referred to a supporter from within Berkshire NHS Foundation Trust. Primary outcome measures will be the Insomnia Severity Index (ISI) and sleep efficiency (SE) as calculated by the programme’s in-built sleep diary. Secondary outcome measures will be the Patient Health Questionnaire-9 (PHQ-9), Generalized Anxiety Disorder-7 (GAD-7), and the Work and Social Adjustment Scale (WSAS) as part of the minimum data set administered routinely by this IAPT site. All measures will be administered at baseline and weekly thereafter.
Wogan, R., Enrique, A., Adegoke, A., Earley, C., Sollesse, S., Gale, S., Chellingsworth, M., Richards, D. (2021). Internet-delivered CBT intervention (Space for Sleep) for sleep disorder in a routine care setting: Results from an open pilot study. Internet Interventions, 26. Link to Paper
The Effectiveness & Cost-Effectiveness of Internet-Delivered Interventions for Depression and Anxiety Disorders in the Improving Access to Psychological Therapies Programme D-IAPT.
For many years now, the UK IAPT services have been using SilverCloud to deliver low-intensity online delivered cognitive-behavioural therapy (CBT) to clients. Data reported nationally through this service demonstrates that SilverCloud interventions perform very well with 70% of clients showing reliable improvements in their symptoms after treatment. The current trial seeks to establish very robustly and independently the effectiveness of online CBT as a treatment for depression and anxiety. It also seeks to show that providing this service is cost-effective for the National Health Service.
Assessing the efficacy and acceptability of an internet-delivered intervention for resilience among college students: A pilot randomised control trial
A pilot randomised controlled trial included three groups: 1) an intervention group with human support; 2) an intervention group with automated support, and 3) a waiting list control group. The intervention, Space for Resilience, is based on positive psychology and consists of seven modules, delivered over a period of eight weeks. Primary outcome measures will include the Connor-Davidson Resilience Scale (CD-RISC) and the Pemberton Happiness Index (PHI). Secondary outcomes measures will include the Brief Resilience Scale (BRS), the Patient Health Questionnaire – 4 items (PHQ-4), the Rosenberg Self-Esteem Scale (RSES), and the Perceived Stress Scale – 4 items (PSS-4). Acceptability will be examined using the Satisfaction with Treatment (SAT) questionnaire. The analysis will be conducted on an intention-to-treat basis.
Enrique, A., Mooney, O., Salamanca-Sanabria, A., Lee, C.T., Farrell, S., Richards, D. (2020). Assessing the efficacy and acceptability of a web-based intervention for resilience among college students: Pilot randomised control trial. JMIR Formative Research. Link to Paper
In Project Talia we are exploring how we can best leverage AI to help improve the effectiveness of important mental health services. This partnership with Microsoft Labs Cambridge and SilverCloud Health aims to jointly explore how AI can be used to enhance SilverCloud Health’s digital mental health services that deliver cognitive-behavioral treatment (CBT) programs to a large and growing number of people in need of effective care. Using probabilistic machine learning frameworks, the aim is to identify new routes for personalizing treatments and improving patient engagement and clinical outcomes.
- Microsoft Research Blog: Data-driven insights for more effective, personalized care in online mental health interventions
- CHI 2020 paper: Understanding Client Support Strategies to Improve Clinical Outcomes in an Online Mental Health Intervention
- Microsoft Future Decoded
- Microsoft Research Blog: Microsoft collaborates with SilverCloud Health to develop AI for improved mental health
Internet-Delivered Cognitive Behavior Therapy as a Prequel to Face-To-Face Therapy for Depression and Anxiety: A Naturalistic Observation
Online CBT has been mostly used with people in the mild-moderate symptom range. Although this is changing and the question that this study was interested in examining was whether we could use the Silvercloud interventions as part of a treatment package offered to clients who were assessed as severely depressed or anxious and waiting for face-to-face treatment. The study also facilitated us taking a look at the important construct of therapeutic alliance online and the perception of the clinical staff of the usability and acceptability of using a digital intervention as part of a package of care for their clients.
ONLINE RECOVERY TOOLKIT FOR BIPOLAR: IRELAND
The current project developed an evidence-based recovery-orientated toolkit as an intervention to support the care of patients with a bipolar diagnosis. We worked with one of the recognised global leaders in Bipolar Disorder, Professor Steve Jones of Lancaster University. The intervention was developed with subject matter expertise alongside stakeholder involvement from patients, service users, and clinicians.
We successfully piloted the intervention at three sites in Ireland. Participants had a mean age of 40 and preliminary results show a significant improvement in measures of recovery and quality of life for patients with bipolar disorder. Overall, the service users showed good levels of acceptability of the intervention, found it easy to use, and they would continue using it over time. They also reported that they found the program quite helpful.
CULTURAL ADAPTATION AND EFFICACY OF ICBT IN COLOMBIA
This study aimed at evaluating the efficacy of a culturally adapted and internet-delivered cognitive behavioural therapy (iCBT) through a randomised controlled trial (RCT) and with 3-months follow-up. The programme consisted of 7-modules. Two hundred and fourteen Colombian college students with depressive symptoms were randomly assigned to either the treatment group (n= 107) or a waiting list control group (n=107). Participants received weekly support from a trained supporter. The primary outcome was symptoms of depression as measured by the Patient Health Questionnaire (PHQ-9) and the secondary outcomes were anxiety symptoms assessed by the Generalised Anxiety Disorder questionnaire (GAD-7).
This trial implemented Silverclouds interventions for depression, anxiety, and stress in a University student population. The aim of the study was to see how well it could be implemented and also assess outcomes. An interesting feature of this trial is the ability of the user to choose which programme they wish to use – a patient preference was included after participants learned their scores on the psychometric measures we used to assess their symptoms on depression, anxiety, and stress levels.