I-Corps: Translation Potential of an Artificial Intelligence (AI) Simulation for Crisis-Ready Behavioral Health Training

Award ID: 2628119

This I-Corps project is based on the development of an artificial intelligence (AI)-enabled behavioral health simulation training platform that provides interactive practice for counseling, psychology, social work, peer support, and crisis-response training. Current training methods rely heavily on role-play and manual supervision, which are time-consuming, inconsistent, and unable to reproduce real emotional pressure. As a result, trainees often enter real-world situations underprepared, which can increase stress for both counselors and the individuals they support and may lead to poor outcomes in high-risk scenarios. These limitations affect training programs across higher education, professional certification, continuing education, and workforce development, contributing to inconsistent skill preparation, rising supervision demands, and insufficient readiness for emotionally intense situations amid persistent workforce shortages. This technology pairs adaptive virtual clients with structured feedback to deliver scalable, repeatable practice that supplements traditional role-play and supervision. By allowing learners to rehearse both routine and high-stakes interactions in a safe environment, the technology may improve workforce readiness, expand access to quality training, and strengthen behavioral health services. This I-Corps project utilizes experiential learning coupled with first-hand investigation of the industry ecosystem to assess the translation potential of an Artificial Intelligence (AI)-powered behavioral health simulation training platform. The technology uses personalized large language models, parameter-efficient model personalization, and competency-aligned evaluation methods to generate realistic client interactions and structured feedback. It integrates generative AI, efficient personalization methods, adaptive conversational modeling, and rubric-based assessment to deliver feedback aligned with behavioral health competencies. The platform supports scenarios ranging from foundational communication skills to crisis intervention and emotionally complex conversations. Its central technical advance is the ability to model individualized client personas whose emotional states evolve across multi-turn conversations, including escalation, resistance, and de-escalation, so the system simulates a responsive person rather than a scripted dialogue. Unlike conventional role-play that depends on faculty availability, standardized patients, or infrequent simulations, the platform enables on-demand, repeatable practice with adaptive virtual clients. Preliminary technical results demonstrate the feasibility of emotionally realistic interactions, repeated scenario practice, and structured evaluations that can both complement and enhance conventional human-based supervision. This may enable a more behaviorally realistic and scalable training environment that may improve behavior workforce readiness and client outcomes. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Program Director: Ruth Shuman