Use of a Generative AI (Gen-AI) Chatbot for Treating Anxiety and Depression Among Persons with Cannabis Use Disorder (CUD)

Cannabis use disorder (CUD) is widely prevalent, a growing public health concern, and highly comorbid with anxiety and depression. Effective cognitive and behavioral treatments exist for co-occurring CUD, depression, and anxiety, but they are inaccessible to many due to financial burden and resource limitations. Thus, we emphasize the great potential benefit of a mobile digital intervention aimed at treating co-occurring depression and anxiety among persons with CUD. Digital interventions powered with generative artificial intelligence (Gen-AI) have shown tremendous promise in building personalized, on demand support at scale.

The proposed P30 project will build on our generative-AI-driven mobile app, Therabot, which our research group has spent over 100,000 hours in the last five years developing for mental health treatment. Promising results from our randomized controlled trial (N = 210) demonstrated high anxiety and depression symptom reduction compared to control in addition to high engagement with an average of 600 messages sent and 6 hours of engagement per user (Heinz et al., 2024). However, it is currently unknown whether Therabot can be used to treat co-occurring substance use disorders, including CUD. Thus, the primary aim of this P30 project involves working directly with persons of lived experiences (i.e., those who have MDD/GAD and CUD) to iteratively refine and apply our Gen-AI app to populations with these co-occurring disorders.

We will recruit 15 participants with co-occurring CUD and depression or anxiety to participate in 3 rounds of interviews, with the ultimate goal of developing the app alongside persons having lived experience. The first interview will focus on project introduction, user attitudes toward gen-AI in therapy, and core app features, including app notifications and user-interface elements. The second interview will focus on participants’ preferences regarding empirically validated content modalities to be included, specifically which content modalities participants believe would be most useful delivered via the on-demand, mobile format. After the second interview, we will work with our app developers to implement feasible changes based on participant feedback, and then participants will participate in a 4-week trial with our adapted Gen-AI, Therabot-CALM (Cannabis, Anxiety, Low Mood). After the clinical trial, participants will return for the third interview and provide both quantitative and qualitative feedback on their experience with Therabot-CALM. Completion of this P30 project will result in the development and preliminary evaluation of Therabot-CALM, a refined version of Therabot accommodating the needs of persons with co-occurring CUD and depression or anxiety. The results from this project will inform the development of a fully-powered randomized controlled trial to test the effectiveness of Therabot-CALM.