The ExTRAPPOLATE project aimed to co-create a computer-based tool to enable mental health practitioners to reflect on their own practice. The project involved using natural language processing and machine learning to analyse the language in psychotherapy transcripts. Stakeholders were involved throughout the project to understand concerns about the approach and help shape the project.

Project summary

It is difficult for therapists to know how well they are performing in therapy sessions. The ExTRAPPOLATE project aims to use natural language processing and machine learning to develop a tool that therapists can use to look back at their own sessions. With this tool, therapists can see what types of interactions are taking place, and how engaged the patient is. This should give them greater insight into how they are working and allow them to improve their technique. The interdisciplinary project team worked together to develop the design for the tool using knowledge from linguistics, clinical psychology, patient experience, digital health and machine learning.

We also organised a series of workshops with patients, carers, psychotherapists, therapy trainers and therapy service managers to explore trust in this system. The workshops raised concerns about the use of machine learning in psychotherapy and obtained feedback on how the tool should be designed in terms of appearance and function, including how the workings of the tool should be explained to the user.

This project was funded by a grant from the EPSRC via the Human Data Interaction Network, ‘Future of Mental Health’.