3 - (IOP 15) Results from the First Randomized Controlled Trial of an Ai-enhanced Cognitive Behavioral Therapy Training Tool

Among low- and middle-income countries, where roughly 85% of the world’s population reside, fewer than 50% of residents with impairing behavioral health disorders receive treatment. As demand grows for mental health services, so too does the recognition of the mental health workforce shortage and clinical skill gaps. In recent years, the digital revolution has begun to penetrate the healthcare industry at large, as applications of Artificial Intelligence across medical specialty areas proliferate. Adoption of these technologies to support the implementation of evidence-based practices are critical to addressing the access and quality issues that plague behavioral healthcare. On-demand approaches to provide standardized, cost-effective, timely, and scalable performance-based feedback of discrete psychotherapeutic competencies are now possible through the use of natural language processing and machine learning. This presentation will re-introduce a responsibly developed Artificial Intelligence (AI)-enhanced tool for training practitioners in cognitive behavioral techniques for psychosis. This tool served as the prototype for subsequent tools targeting motivational interviewing, CBT for depression and anxiety, dialectical behavior therapy, cognitive processing therapy, and culturally responsive care practices. The authors will present learner and client outcome data from a multistate, multisite NIMH-funded Research & Development grant comparing the tool to training as usual among community behavioral health providers and their outpatients. Providers (N=102) were primarily master's level practitioners (46.1%), followed by bachelor’s-level (35.3%), associate’s (6.9%), or some college (4.9%) who reported a mean of 5.4 years experience working with patients with psychosis. Each enrolled provider had between 1 and 10 patients on their caseload enrolled in the study. Patients (N=300) were primarily diagnosed with a schizophrenia spectrum disorder (75.7%), followed by an affective disorder (23.4%). Overall, both non-specific psychotherapeutic competencies and CBT-specific competencies significantly increased pre- to post-training. Inferential analyses are currently underway. We will conduct a nested mixed model analysis to assess the effect of the training on client symptoms, functioning, and recovery. We will apply a NIMH experimental therapeutics approach to examine the hypothesized mechanism by which the training may influence clinical outcomes through increased provider competencies (mediation). Additionally, we will test whether the strength of this relationship is moderated by client skill acquisition. We will discuss the potential applications of AI-enhanced psychotherapy training to address the global shortage of CBT and CBT-informed care.