What is it about?

Therapists often use measurements during psychotherapy to monitor whether clients are improving. Usually, the answers are combined into an overall score. However, individual measurement items may contain useful information that is lost when only the total score is considered. We analysed routinely collected data from 9,591 clients receiving therapy at university counselling centres. Using two artificial-intelligence-based statistical methods, we examined which answers during the first three to five sessions were most useful for predicting whether clients would achieve meaningful improvement by the end of treatment. Five areas were particularly informative: self-esteem, anxiety, substance use affecting work or study, difficulty concentrating, and functioning at work or school.

Featured Image

Why is it important?

Identifying meaningful changes early in psychotherapy may help therapists and clients make more timely and informed treatment decisions. Our findings suggest that measurement-based care systems could provide more useful feedback by highlighting changes in specific measurement items, rather than relying only on a total mental health score. For example, continuing difficulties with self-esteem, nervousness, concentration, substance use, or work and study functioning may indicate areas that need closer attention. The study is a proof of concept, so the findings need to be tested in other settings before they can be used as clinical decision rules. However, they demonstrate how routinely collected client feedback and artificial intelligence may be combined to make progress monitoring more clinically informative.

Perspectives

We wanted to explore whether routine outcome data contain clinically relevant signals that are overlooked when responses are reduced to a single total score. The findings highlight the potential value of looking more closely at what is changing for a client, rather than focusing only on how much the overall score has changed. Different aspects of well-being may change at different stages of therapy, and item-level feedback could help therapists and clients have more focused conversations about progress. Artificial intelligence should not replace clinical judgement and cannot replace the therapeutic relationship. Instead, it may help organize complex information and draw attention to patterns that therapists and clients can consider together.

Annika Helgadóttir Davidsen
University of the Faroe Islands

Read the Original

This page is a summary of: Early identification of psychotherapeutic change: AI-derived predictors from routine data., Journal of Counseling Psychology, July 2026, American Psychological Association (APA),
DOI: 10.1037/cou0000876.
You can read the full text:

Read

Contributors

The following have contributed to this page