Chatbot for Depression, Anxiety, and Eating Disorders

Testing an Artificial Intelligence Chatbot for Mental Health

The research team would like to test how efficacious an artificial intelligence chatbot is in delivering supportive behavioral interventions in populations with anxiety, depression, or eating concerns.

The research team has developed a deep learning-powered application that can provide support and therapeutic suggestions via natural dialogue. The chatbot is designed to be highly scalable such that it could address access issues (to address the majority of persons who do not not have access to therapy). Eventually, it is hoped this intervention can provide accessible therapy for those with anxiety, depression, and eating disorders, among other mental illnesses. The chatbot has been trained with intervention-oriented text conversations written by research assistants, employing a variety of therapeutic techniques, including cognitive-behavioral therapy. The research team hopes to test how well the chatbot can respond and aid those with mental health symptoms. The chatbot's effectiveness will be monitored based on the participants' interactions with the chatbot, as well as the participants' scores on the PHQ-9, GAD-IV, SWED, and general satisfactory questionnaires.

Inclusion Criteria:

• We will recruit participants online using Facebook, Google Ads, Reddit, MTurk, listservs, and other online platforms.
• These individuals must meet the criteria for anxiety, depression or disordered eating, according to the PHQ-9, GADQ-IV, and SWED 3.0.

Exclusion Criteria:

• Individuals currently being treated for active suicidality (measured with C-SSRS), Bipolar I/II, and active psychosis will be excluded from the study.