Fig. 1

The study’s analytic pipeline. A (Left) The primary dataset analyzed, consisting of 450 clients and 2,444 therapy sessions. For each client, we extracted the number of days between two assessments. (Right) For each pair of consecutive sessions, we assessed whether action recommendations given in the former session were reviewed in the latter session’s dialogues and calculated the number of these occurrences & their percentage. B (Left) Using these calculations, we generated a comprehensive table containing all the data from (A) along with the main calculated metric for each session: the “Review Percentage”. (Right) With the complete data in hand, we examined the relationship between the review percentage and the changes in both depression (PHQ-9) and anxiety (GAD-7) during treatment (here, an illustration of PHQ-9 is shown; this figure is an illustrative representation of our findings, rather than a direct display of the actual results)