Hundreds of millions of people are already using Large Language Models (LLMs), including for mental health purposes, which has led to inadvertent harms. Critically, people with mental health conditions may be especially vulnerable to such harms.
In this project, we will develop the first computational framework to systematically quantify and benchmark the risks that LLMs present to people with mental health conditions. Our approach will simulate interactions of hundreds of users and LLMs to evaluate safety across a variety of mental health conditions, demographics, and AI failure modes.
Targeted Condition: General Mental Well-Being
Optimizing telemental health with live artificial intelligence clinical scaffolding and feedback
This project aims to develop a clinical scaffolding system to enhance telemental health care by providing real-time coaching and actionable suggestions during video-based sessions. Modeled after live supervision methodologies, it supports clinicians by identifying intervention targets and offering text-based coaching prompts to guide care. Unlike automated chatbots, this approach enables clinicians to adapt suggestions to patient needs, balancing automation with oversight for safer AI-supported mental healthcare. The proposed in-session support will facilitate efficient implementation of strategies and clinician skill development. This project seeks to enhance data privacy by processing all data on-device and avoiding external data transfers.
Self-directed mindfulness in medically hospitalized patients: a pragmatic trial
Consultation-liaison psychiatrists are often asked to manage mood and anxiety symptoms experienced by patients admitted to medical and surgical floors of the hospital. This study aims to determine the feasibility and effectiveness of a self-directed mindfulness intervention as an adjunctive treatment for improving mood, anxiety, and perceived stress in medically hospitalized patients. Participants, consisting of patients evaluated by the consultation-liaison psychiatry service, will be randomized to an adjunctive mindfulness intervention group or a treatment-as-usual control group. Feasibility and acceptability of implementing a mindfulness intervention will be assessed. Group differences in the changes in symptom severity and psychotropic medication administration will be investigated.
Centering community voices in partnered mixed methods approaches to addressing health disparities with diverse communities
Meaningful partnerships between community and academic/public health co-researchers are needed to address community mental health, wellbeing, and health disparities. Partnerships should center the strengths, needs, and interests of communities on the communities’ terms while building relationships on trust and shared power.
UW researchers in partnership with the Community Health Board Coalition (CHBC) have developed an initial draft of a Community-Centered Approaches to Research and Evaluation (CCARE) Toolkit to guide academic, public health, and community researchers in partnering to act on emerging community data during times of crisis and beyond. As a part of this work, we are striving to better understand and support community health priorities.
Willow Study
The goal of this study is to understand the impact of stigma on mental health and recovery from trauma in different parts of the country.
Coach up the coaches: extending the reach of mental health professionals in sport settings
More than half of school-aged youth in the US play at least one organized sport, are similarly at-risk for mental health disorders as compared to their non-athlete peers, and face heightened barriers to seeking and sustaining mental health care. While coaches are not licensed mental healthcare providers, there is a robust evidence-base about laypeople (e.g., parents, teachers) effectively delivering brief interventions across population settings to help reduce psychological distress and increase the adoption of health behaviors. However, such brief interventions have not as-yet been adapted for the sport setting, or delivered by coaches.
The overarching goal of this project is to identify sport setting appropriate (very) brief interventions and adapt them for coach delivery in the sport setting. This will be accomplished by working in partnership with coaches, athletes, and licensed mental health care providers. We will use University of Washington’s First Approach Skills Training (FAST) training model (originally designed to train primary care clinicians in mental healthcare skills) to train a pilot cohort of coaches to deliver these interventions. This project is the starting point for a potentially transformative opportunity to extend the reach of mental healthcare professionals into the large and often underserved population of youth sport participants.
Partnering with the Quinault Indian Nation to develop a community-based contingency management intervention
Indigenous knowledge remains strong in American Indian/Alaska Native (AI/AN) communities despite colonization, postcolonial efforts to assimilate and eradicate AI/AN people, present-day inequities and structural racism. Despite strength and resilience, AI/AN populations have high rates of unintentional overdose and death by suicide. There is an urgent need for evidence-based, culturally responsive approaches to increase protective factors against substance use and mental illness in AI/AN communities.
This study is a partnership between University of Washington researchers and the Quinault Indian Nation’s behavioral health providers and community. We will design a culturally adapted community-based contingency management intervention to increase positive reinforcement for adolescents and young adults at risk for substance use and mental illness. This intervention will focus on community identified target behaviors rather than abstinence. By prioritizing the needs of the community we will develop an intervention that is effective, usable and sustainable.
Family and Caregiver Training and Support Program (FACTS) pilot
We know from decades of research that caregiver involvement, including family and non‐family members, in a patient’s mental health treatment can make a tremendous difference in the trajectory of their loved one’s life by supporting recovery, reducing relapse, and decreasing mental health crises. Family and caregiver involvement also decreases provider stress, improves caregiver well-being, and can lead to lower patient healthcare utilization and costs. But despite their importance, many family members and caregivers struggle to engage in the kind of support that can benefit the patient and themselves. They often lack access to education, resources, or skills to step into this critical role despite a desire to help. Our initiative intends to develop a pilot Family and Caregiver Training and Support Program (FACTS) program that aims to decrease barriers to caregiver involvement and improve caregiver support.
Our team will develop online training that will include an orientation to having a loved one who is psychiatrically hospitalized and will teach families and caregivers practical communication skills while their loved one is in our care. These topics would be relevant regardless of a patient’s diagnosis and will be adapted from existing evidence‐based models. The pilot will be tested with caregivers of patients hospitalized at the new Center for Behavioral Health and Learning and we will proactively integrate input and feedback from participants to inform program improvements along the way.
We will also build a public-facing website to host FACTS training materials as well as mental health information and resources that we will curate for accuracy and reliability. We expect the FACTS pilot content will serve as a foundation for additional offerings that will include diagnosis specific skills trainings as well opportunities for in-person sessions and Family Peer Support programming.
Implementing contingency management in a supportive housing context to promote resident well-being
Mental health and substance use disorders carry complex needs of living that are often poorly-served by our health systems, which result in immense societal costs. This has prompted federal recognition of supportive housing among the vital conditions that support the long-term recovery and resilience of our citizenry. Supportive housing offers safe living environs where residents access services such as harm reduction supplies and referral to recovery-oriented activities like peer support groups and workplace re-entry programs. To maximize the benefit that residents experience from such services, behavioral health staff working in supportive housing contexts should implement useful psychosocial therapies to spur greater service utilization by residents.
One such therapy that has proven useful among persons with mental health and substance use disorders is contingency management (CM), in which one earns tangible rewards for demonstrating desired behaviors like attending a support group, taking a prescribed medication or completing workplace re-entry paperwork. Given the robust and reliable therapeutic impacts of CM, there is much to be gained from its implementation in supportive housing environs. For several years, the project lead Dr. Hartzler and his team at the UW Center for Advancing Addiction Health Services (CAAHS) have provided comprehensive technical assistance to regional treatment settings to facilitate design and successful implementation of sustainable CM programming. Also, UW project collaborators Drs. Fockele and Leyde have built working relationships with Downtown Emergency Services Center (DESC) while exploring interest in CM among its staff and residents. This convergence of acute community need, an experienced team offering technical assistance and an established academic-community partnership highlights an opportunity in which CM will be implemented in DESC-governed supportive housing facilities to increase resident utilization of beneficial services.
A pilot trial on EMA habit formation behavioral strategies for improving engagement of digital mindfulness interventions among non-suicidal self-injury engagers
Non-suicidal self-injury (NSSI), the purposeful, direct damage of one’s body without the intent to die, is a pervasive public health concern with clinically significant long-term consequences. Mindfulness – a core skill in DBT, an evidence-based treatment for NSSI, is designed to target emotion dysregulation and rumination and may be particularly relevant due to the proliferation of digital mindfulness interventions in recent years. To this end, the goal of this study is to expand the use of ecological momentary assessment (EMA) and to develop and evaluate a program of habit-formation strategies (e.g., SMART-goal setting, reinforcement scheduling) to boost user engagement and treatment effects of DMI. Following a 1-week EMA baseline period, participants (N=40) will be randomized to either TAU (Mindfulness only) or Experimental (Mindfulness + Behavioral Prompts) conditions for a 4-week intervention EMA period.
