Lehigh professors receive grant to help address mental health care language gap with AI training tool

At a time when many people seeking mental health care still struggle to find a clinician who understands their language and culture, two Lehigh University researchers, along with a third from George Mason University, have received grant support from Lehigh University for a project aimed at helping train providers to close that gap. 

Vanesa Mora Ringle, assistant professor in Lehigh University’s College of Education, along with co-principal investigators Dr. Mooi Choo Chuah, professor in Lehigh University’s P.C. Rossin College of Engineering and Applied Science and Dr. Amanda Sanchez, assistant professor in George Mason University’s College of Humanities and Social Sciences, is leading the CALLM-Care project, short for Culturally Adaptive LLM-Guided Care. The researchers received a grant from Lehigh’s OVPR Fund to Integrate Artificial Intelligence (AI) in Research to support the project. 

CALLM-Care is being developed as a training tool for mental health therapists, not a replacement for them. Using artificial intelligence, the platform will simulate patients from different cultural and linguistic backgrounds, allowing trainees and practitioners to practice counseling interactions and receive feedback on their cultural and linguistic responsiveness. The system includes a client simulator, a therapist module, and a supervisor module that provides multidimensional feedback. The project is rooted in a clear problem within the mental health workforce. People who speak non-dominant languages often face barriers to care because too few providers are prepared to serve them in linguistically and culturally responsive ways. The work also addresses the long-standing challenge of English-centric training models that can limit how bilingual clinicians use their language skills in practice. 

“Mental health access does not begin when a patient gets an appointment,” Mora Ringle said. “It begins when that patient can explain pain, fear, trauma or family pressure in the language they prefer and know the person on the other side will understand more than just the words.”

The team will build a synthetic dataset with a stronger cultural context, fine-tune the platform to simulate patients from varied backgrounds and languages, and test the system with 30 mental health providers, including monolingual and bilingual participants, over a four-week period. Feedback will focus on ease of use, cultural and linguistic authenticity, and other training outcomes.

“Too often, language difference is treated as a problem to work around instead of a strength to build on,” Mora Ringle said. “We need to train more bilingual clinicians and help them use their language skills as clinical assets, because communication is the foundation of treatment.”

The work builds on progress already underway. The team has developed an initial CALLM-Care platform and is soon launching a pre-pilot study involving 20 therapy trainees, including bilingual English-Spanish and monolingual English participants. The current phase is designed to strengthen the platform, generate proof-of-concept data, and support a later application for a National Institute of Mental Health planning grant.

“Artificial intelligence may help improve training, but it has to be evaluated carefully for bias, ethics, and cultural fit,” Mora Ringle said. “If a tool is meant to improve cultural responsiveness, it has to be culturally responsive itself. Otherwise, it risks reinforcing the very gaps it is supposed to close.”

The $12,926 grant supports this essential project, which also creates opportunities for cross-disciplinary student research in computer science and counseling psychology, bringing technical development and clinical expertise together in a shared effort. 

“This is not just a technology project,” Mora Ringle said. “It is a test of whether higher education can help build a mental health workforce that better reflects the language and culture of the people it serves.”