A new study published in the Frontiers of Health Services looked at leveraging generative AI to simulate mental healthcare access and utilization.
“This study is about adapting AI to improve efficiency and effectiveness of mental health services,” study author Cortney VanHook told us. VanHook is an assistant professor of social work at the University of Illinois Urbana-Champaign. “I set about testing if AI software could generate a robust and culturally sound assessment and treatment plan for a fictitious client.”
Researchers see an opportunity to integrate the computing power of AI with clinical know-how to understand populations being served.
“The truth is that clinical skills and the therapeutic alliance are not sufficient to achieve positive treatment outcomes,” VanHook told us. “Our clients are operating at the level of a first or second-generation large language model. They are extremely complex when compared to individuals seeking mental health services five, ten, and 20 years prior.”
It is also true that readiness to change is not sufficient, explained VanHook.
“As clinicians we all know readiness to change is quite fickle and clients can easily swing towards despair when they encounter setbacks,” VanHook told us. “We have to identify and be responsive to the psychosocial and structural barriers clients experience when attempting to interact with mental health services. In this article you see that we apply both a prevention and intervention lens in our approach.”
VanHook chose this topic because she believes that we do not sufficiently engage students and junior clinicians in thinking about their work in a structural manner.
“As clinicians, we are beholden to all of the ecological factors in our clients’ lives,” VanHook told us. “This article provides guidance to strengthen clinicians’ structural competency.”
AI outputs are not without flaws, however, VanHook explained. AI is only valuable when it reflects clinical experiences and the scientific literature. AI should be used to triangulate data, not to lead in treatment approach.
“I hope administrators and managers of mental health programs access this article,” VanHook told us. « If this approach is successful, it will be because leadership validate and implement aspects of our approach.”
VanHook would like to see clinicians, service leaders, and researchers utilize this method for different populations including rural, immigrants, refugees, and individuals with intellectual disabilities.
“Publish your results,” VanHook told us. “Publication is not solely the domain of PhDs. »