A unified protocol chatbot reduces anxiety by encouraging university students’ negative emotional expressions: A randomized controlled trial

The use of chatbot services has been rising in the field of mental health; however, the therapeutic mechanism underlying them remains unclear. Hence, through a randomized controlled trial, we aimed to show that a chatbot utilizing a Unified Protocol (UP) for the transdiagnostic treatment of emotional disorders—referred to as the UP chatbot—could reduce symptoms of anxiety in university students by encouraging their negative emotional expressions. We randomly assigned 310 Japanese university students to one of four groups: one that received only the UP chatbot service; one that received only the UP guidance service; one that received both services; and a waitlist group that did not receive any service. Over an 8-week intervention and a 4-week follow-up, the participants filled out questionnaires assessing symptoms of anxiety and depression. Additionally, we employed sentiment analysis to examine their emotional expressions with the chatbot. Our analysis of a latent growth curve model indicated that the participants in the group receiving only the UP chatbot service reported significantly higher reductions in symptoms of anxiety than the waitlist group; these effects were sustained during the follow-up. Furthermore, they demonstrated significantly stronger expressions of sadness than the waitlist group. Thus, the UP chatbot, similar to the UP human therapist, may have encouraged the participants to express negative emotions, facilitated emotional acceptance, and reduced their anxiety. • A randomized controlled trial tested a chatbot’s impact on students’ anxiety. • The chatbot incorporated interventions based on a unified protocol. • The chat-only group experienced reduced anxiety compared to the waitlist group. • The chat-only group expressed more instances of sadness than the waitlist group. • The chatbot, like the human therapist, facilitated students’ expressions of sadness.