Back to top

In-Session Psychological Therapy Feedback

Live: ARC Theme 

Developing in-session prognostic feedback during person-centred counselling and cognitive behavioural therapy for depression

Why the research is needed

Depression is a common and costly mental health problem which can become a long-term issue. Psychological therapies including Cognitive Behavioural Therapy (CBT) and Person-Centred Experiential Therapy (PCET) are recommended, effective treatments. However, psychological therapists themselves can vary considerably in how effective they are for their patients. For example, the most effective therapists can have rates of improvement 10 times the average, whilst the least effective therapists can make their patients’ problems worse. Furthermore, therapists do not become more effective with time or experience and differences between therapists become more pronounced in short-term psychological therapy – the predominant modality offered in NHS services. Overall, this means that patients cannot rely on the general effectiveness of a psychological therapy because of the potentially large differences between they may see. For therapists this means that they cannot rely on time or experience to help them improve.

What is already known about the subject

Offering feedback to psychological therapists on patient prognosis during therapy can help improve effectiveness, and more detailed feedback increases this effect. This means using known indicators of whether treatment is likely to help or not. Currently, prognostic feedback is limited to self-report questionnaire scores, but this study could help extend feedback to include details of what helps and what hinders from the content of psychological therapy sessions.

Who we will be working with

  • Grazziela Figueredo and Tom Trimble from the University of Nottingham Digital Research Service will support the machine learning processes.
  • Jeremie Clos from the University of Nottingham School of Computer Science will contribute to the Natural Language Processing methods used and support ethical and trustworthy approaches to using AI.
  • The study uses data from a completed clinical trial run at the University of Sheffield. Dave Saxon was the statistician for the trial and Gillian Hardy one of the project leads. Both will be contributing to the project and enabling data access.
  • Nima Moghaddam from the University of Lincoln will contribute expertise on psychological therapy process research.

How patients and the public are involved

Two patient advisors are members of the research team so that patient perspectives shape all research decisions made. An additional group of public contributors will help shape the study’s subsequent software development, to support explainability and trust among patients and practitioners. Their contributions will also help to reduce the bias against minority and socially disadvantaged groups often present in AI applications.

What we will do

We will transcribe CBT and PCET sessions and rate them using a tool that can give indications of what is helpful and what is unhelpful in therapy session contents. We will use types of artificial intelligence called machine learning and natural language processing to automate this rating process. We will then see if this automated rating tool can give helpful indicators of prognosis across therapy.

What the benefits will be

We will gain a better understanding of the most helpful and hindering aspects of CBT and PCET may be. We will also be able to identify key trainable behaviours that can help therapists improve their effectiveness.

When the findings will be available

Initial findings will be available 31st March 2023, but development is planned to continue until August 2024 after this project is completed.

How we are planning for implementation

The algorithm developed in this project will be incorporated into a piece of software that will be given to a group of NHS psychological therapists in a feasibility study that will follow on from the current work. This will help us understand how this information can be used in clinical practice.

Contact

Sam Malins, sam.malins@nottshc.nhs.uk / sam.malins@nottingham.ac.uk