Chatbot-delivered acceptance and commitment therapy with adolescents: A pilot randomized controlled trial

Abstract

To date, no published randomized controlled trial (RCT) has evaluated chatbot-delivered acceptance and commitment therapy (ACT) for youth mental health. The present pilot study was a three-arm parallel-group RCT that aimed to (i) evaluate universal chatbot-delivered ACT for adolescents and (ii) assess the feasibility of nesting high temporal density measures of problematic rule-following (i.e., daily ecological momentary assessments) in the chatbot. 148 young people aged 15 to 17 years were randomly assigned to receive either chatbot-delivered ACT, psychoeducation via a booklet, or neither of these (non-active control). Participants completed self-report measures of generalized pliance, tracking, affect, psychological inflexibility, and valued living at three time-points (immediately pre-intervention, immediately post-intervention, and at a six-week follow-up), with participants assigned to the chatbot further completing daily single-item assessments of problematic pliance. No statistically significant changes in any dependent variable were observed across conditions or over time. As per intervention satisfaction data, chatbot-delivered ACT demonstrated adequate acceptability. Given both the apparent convergence with a standardized measure and minimal missing data, nesting ecological momentary assessments in the chatbot was deemed highly feasible. Findings are discussed with explicit reference to (i) appropriate session duration and (ii) means of enhancing artificial intelligence and natural language processing technologies underlying the chatbot going forward.