New $12.6M grant to help unravel hallucinations

Researchers in psychiatry and biomedical informatics create a predictive AI tool to get people appropriately into care.

Media Contact: Chris Talbott - 206-543-7129,

UW Medicine researchers have received a five-year, $12.6 million grant from the National Institute of Mental Health to create a digital evaluation tool for people who experience hallucinations.

Hallucinations are the most common psychotic experience. They occur in about 13% of the U.S. population and are associated with a range of conditions, from substance use to head trauma and schizophrenia. 

The new project is meant to help clinicians identify the severity of hallucinations and the need for a deeper level of care.

“There just isn't the capacity in clinics or in hospitals to be able to assess the large number of people that experience hallucinations,” said Dror Ben-Zeev, professor of psychiatry and behavioral sciences at the University of Washington School of Medicine. “You have to conduct a comprehensive clinical interview and monitor people over time to determine who actually needs intensive services and who is otherwise healthy. 

“The technologies that we have for clinical interviews, they're not up for the task. Here we’re trying to deploy a combination of different measures on a smartphone platform that might provide digital indicators of when someone might require time-sensitive care.”

Ben-Zeev and Dr. Trevor Cohen, professor of biomedical informatics and medical education, will serve as principal investigators on the grant. Both have a background in applying clinical experience to inform technology design.

The point of the app is to identify the severity of hallucinations and the need for treatment, Ben-Zeev said, not to be used as a diagnostic tool.

“This is a relatively new direction in mental health research where it's really not meant to focus on diagnoses because we know that hallucinations occur in a lot of different conditions,” said Ben-Zeev, a clinical psychologist and director of UW Medicine’s Behavioral Research in Technology and Engineering Center. “We're trying to focus in on the experience of hallucinations, both hearing voices and seeing things, as a standalone, clinically meaningful experience giving us an indication of brain functioning.”

To do this, researchers will gather three streams of information using a smartphone platform, Cohen said. They’ll take momentary assessments of how users are doing on a rating scale. The investigators also will collect speech data from users’ audio diaries that describe the experience of hallucinations. And participants will be asked to remember a story to allow researchers to collect more speech data and indicators of cognitive functioning. 

The grant will include 2,000 participants, thereby making it the largest data-collection study of its kind, the researchers said. The results will be fed into a predictive modelling program.

“We're going to apply AI methods, including methods from the same family as ChatGPT, to extract features from the language data,” Cohen said. “All of those extracted features will be part of predictive models to try to detect when people will deteriorate.”

The project involves a large team at UW and collaborators at the University of Minnesota and Louisiana State University. 

The project will be funded by NIMH grant U01MH135901.

For details about UW Medicine, please visit

Tags:behavioral health

UW Medicine