Researchers from the University of Pennsylvania, New York University, and Penn's Linguistic Data Consortium announced on July 9 that they have received a two-year, $4 million grant from the Wellcome Trust to develop an AI-powered platform aimed at training mental health clinicians.
The project, named STELLAR (Steering-Vector Enhanced LLM Agents for Realistic Digital Twins in Mental Health), will create virtual patients known as "digital twins." These AI-driven simulations are designed to allow trainees to practice clinical interviews with patient profiles whose psychiatric symptoms can be precisely adjusted. The goal is to provide an ethical and repeatable way for trainees to simulate interviewing patients across a range of symptom presentations, backgrounds, and clinical scenarios.
Sharath Chandra Guntuku, Research Associate Professor in Computer and Information Science within Penn Engineering and one of the project leads, said, "STELLAR brings together behavioral data, clinical expertise and AI to ask a very practical question. Can we build training tools that better prepare clinicians for how varied and complex patients are?" João Sedoc, Assistant Professor of Technology, Operations and Statistics at NYU's Stern School of Business and another project lead, added, "If we can create digital patients that simulate controllable plausible symptom expression and responsibly evaluate, we can augment current clinician training practices with the kinds of conversations that are essential to better mental health care."
The simulations will be based on composites drawn from real-world data rather than copies of individual patients. Clinical data from the Philadelphia Neurodevelopmental Cohort—a resource founded by Penn Medicine and the Children's Hospital of Philadelphia—will inform these simulations. The project will also use data from social media platforms where mental health symptoms may appear in everyday language. Guntuku noted, "Many mental health symptoms do not appear only in formal clinical settings; they also come through in the way people talk day to day, including online."
Ryant said, "LDC's role is to bring speech and language science into the core of the project: adapting speech-recognition tools to clinical interviews, creating high-quality transcripts and annotations, and helping evaluate both what the simulations say and how they say it." He added this includes assessing model-generated language quality as well as evaluating synthetic voices' naturalness during trainee interactions.
People with lived experience of mental health conditions—as well as family members or caregivers—will be involved throughout development. Gur said, "By involving individuals with lived experience throughout the project, STELLAR can help us ask not only whether a digital patient is clinically accurate but whether the interaction feels respectful, realistic and attentive to experiences that are too often missed." Guntuku concluded, "By connecting real-world symptom expression with controllable digital simulations, we hope to make clinician training more scalable, rigorous and representative."
