Deep learning could keep CBT therapists on track, says study

A study has shown that deep learning could be used to monitor therapists’ effectiveness at conducting cognitive behavioural therapy (CBT), a favoured method for dealing with mental health problems such as depression.

It’s the latest example of the technology’s use at gaining clinical insights from large amounts of data that would previously have taken vast amounts of time and effort from human researchers.

Authors of the UK-based research published in JAMA Psychiatry noted that detailed monitoring of therapist performance requires expensive and time-consuming procedures.

The research led by Michael Ewbank of the Clinical Science Laboratory in Cambridge, said the monitoring could help to stop “therapist drift”, where practitioners begin to lose sight of the techniques they learnt during their training.

The data set based on transcripts from CBT sessions represented 90,000 hours of treatment, more than three times the amount a typical therapist would accrue during their careers.

Data were obtained from patients receiving internet-enabled CBT for a mental health disorder between June 2012 and March 2018 via a commercial package used in England’s NHS.

Clinical outcomes were measured according to whether a patient was engaged and reliable improvement was calculated based on standard mental health questionnaires corresponding to depression and anxiety symptoms.

Using deep learning techniques, researchers analysed transcripts therapy features, such as mood checks, updates, goal setting and homework review.

Results showed that increased quantities of these therapy features were associated with greater odds of reliable improvement.

In contrast, increase in non-therapy related content, such as greeting or saying goodbye, or “other” content was negatively associated with improvement.

Researchers conceded that the technique cannot yet determine whether a therapist adheres to a CBT protocol.

But future work could build on this approach to generate a validated method of assessing session quality and adherence to CBT guidelines.

Authors wrote: “Such monitoring could help arrest therapist drift, i.e. the failure to deliver treatments a therapist has been trained to deliver, which may be one of the biggest factors contributing to poor delivery of treatment.

“ As such, we believe this approach represents an important step in developing a data-driven understanding of mental health treatment and in improving the efficacy of psychotherapy.”