A Case for Leveraging Generative AI to Expand and Enhance Training in the Provision of Mental Health Services (Preprint)

2026-01-09

<sec> <title>UNSTRUCTURED</title> Generative artificial intelligence (Generative AI) is transforming healthcare. With this evolution comes optimism regarding the impact it will have on mental health, as well as concern regarding the risks that come with generative AI operating in the mental health domain. Much of the investment in, and academic and public discourse about, AI-powered solutions for mental health has focused on therapist chatbots. Despite the common assumption that chatbots will be the most impactful application of GenAI to mental health, we make the case here for a lower-risk, high impact use case: leveraging generative AI to enhance and scale training in mental health service provision. We highlight key benefits of using generative AI to help train people to provide mental health services and present a real-world case study in which generative AI improved the training of veterans to support one another’s mental health. With numerous potential applications of generative AI in mental health, we illustrate why we should invest in using generative AI to support training people in mental health service provision. </sec>

CREATE Curated Article List

OSF Preprints (OSF Preprints) · 2026-01-15

***Click on the "Files" tab at the top to see papers, organized into folders by topic**** We will add publicly available papers and links to preprints for papers related to the USE of LLMs for mental health and implementation here.

Psychometric properties of the Oppression-Based Traumatic Stress Inventory and measurement equivalence across PTSD treatment and diverse undergraduate samples.

Psychological Trauma Theory Research Practice and Policy · 2026-01-22

OBJECTIVE: Research demonstrates that oppression can produce symptoms consistent with posttraumatic stress disorder (PTSD), but traditional trauma assessments do not account for the impacts of oppression. This study addressed this gap by establishing the dimensionality, measurement equivalence, reliability, and convergent validity of the Oppression-Based Traumatic Stress Inventory across two samples. METHOD: = 227) who completed a series of questionnaires, including the Oppression-Based Traumatic Stress Inventory. RESULTS: ) that yielded acceptable fit after adding four error covariances. Measurement invariance testing revealed three of the 25 items had parameters that differed across samples. Excellent reliability was found for all three factors. A higher order factor appeared plausible but was largely noninvariant across samples. Finally, we provide evidence for convergent validity (with measures of standard PTSD, posttraumatic cognitions, depressive symptoms, psychosocial functioning, racial discrimination, gender discrimination, and, to some degree, material hardship). CONCLUSIONS: Our findings strengthen the psychometric evidence supporting this novel measure of oppression-based traumatic stress, an important step in furthering intersectional research on this topic. (PsycInfo Database Record (c) 2026 APA, all rights reserved).

A Framework for Evidence-Based Psychotherapy with AI (EBP-AI)

PsyArXiv (OSF Preprints) · 2026-04-29

Artificial intelligence (AI) systems and large language models (LLMs) offer substantial potential to augment or even fundamentally change elements of psychological assessment and treatment. However, current AI technologies have yet to demonstrate the capacity to effect meaningful and sustained clinical change. This gap reflects both the limited integration of clinical science knowledge into language models and applications built using them, as well as the mismatch between the brief, minutes-long nature of most AI interactions and the months-long course of most evidence-based treatments. Here we introduce the Evidence-Based Psychotherapy with AI (EBP-AI) framework, which articulates a set of principles for developing effective clinical AI applications: 1) psychodiagnostic assessment, 2) longitudinal case conceptualization, 3) appropriately dosed intervention planning, 4) meaningful progress evaluation, 5) rigorous validation with clinical populations, 6) attention to real world implementation and use, 7) clinically appropriate style, and 8) understanding clinical psychology as a living science. We introduce a set of key technical questions for the development and evaluation of clinical LLMs and AIs aligned with these principles. Despite their potential, current clinical AIs fall short, in part due to issues with memory, sycophancy, and prioritizing short-term helpfulness over long-term clinical impact. Responsible and ethical design of effective, clinical-science-based AI systems will require understanding their limitations and strategically extending their capabilities.

Unpacking post-implementation adaptations to peer support for adults with mental health challenges: qualitative findings

Mental Health and Social Inclusion · 2026-05-08

Purpose The integration of paid peer support workers into publicly funded mental health services has gained momentum over recent decades. While research has primarily focused on the effectiveness and early implementation of peer support, limited attention has been paid to how these services evolve and are sustained over time. This paper aims to report on adaptations made to a peer support service following its initial implementation and during its sustainment phase. Although such changes are often necessary throughout the implementation cycle, they remain under-documented, leaving gaps in our understanding of their impact on implementation outcomes, including long-term sustainment. Design/methodology/approach Drawing on qualitative methods, 19 individuals were interviewed. Participants were engaged in implementing and sustaining peer support in an outpatient unit overseeing community housing for adults with mental health challenges. Data collection and analysis were guided by the Framework for Reporting Adaptations and Modifications-Enhanced (FRAME). Findings Participants described seven key modifications made across three levels; micro, meso and macro adaptations based on where they originated and their perceived impact. These were changes to: service delivery mode, peer support workforce, referral processes, support structures, service expansion and peer worker tasks. Most adaptations emerged organically in response to evolving needs, rather than through formal planning. Importantly, none compromised the core principles of peer support, though all had a multimodal impact on the peer support service. Originality/value This study underscores the significance of adaptation in sustaining peer support services. It illustrates the diverse range of post-implementation changes (planned and emergent) and their impact on sustainment. We offer practical guidance for service leaders and practitioners on how to systematically document and assess adaptations over time. By showing how changes can occur without compromising core peer support principles, the analysis supports more responsive and sustainable implementation. Additionally, it demonstrates how FRAME can be applied in real-world settings to guide decision-making during the sustainment phase and highlights the need to examine how specific adaptations influence implementation and sustainment outcomes.

Conversational AI Should Fill the White Space in Mental Health Care, Not Replace Humans

2026-03-23

Conversational AI (CAI) is being rapidly deployed in mental healthcare settings, with much of the attention focused on using chatbots to deliver therapy directly to patients. However, we argue that this vision is too narrow and overlooks broader opportunities for CAI to extend existing systems of care. Mental health care systems have struggled for decades with structural gaps outside of immediate therapy delivery. Specifically, we identify opportunities for CAI to support (1) prevention and screening, (2) patients on waitlists, (3) between-session processes during ongoing therapy, and (4) post-treatment aftercare. Patients spend most of their time in these stages where human support is currently the least available. Rather than focusing on replacing clinicians in therapy, we argue that CAI should be strategically integrated into the broader care infrastructure to make the care system more seamless, less burdensome, and increase its capacity, thereby enhancing its effectiveness. We review preliminary studies that suggest that this is plausible. Across stages of healthcare delivery, we highlight specific opportunities and candidate mechanisms for effective and ethical implementation of such blended human-AI care.

Generative Artificial Intelligence in PTSD Treatment: Exploring Five Different Use Cases

2025-05-15

Posttraumatic stress disorder (PTSD) is a prevalent and debilitating condition, yet many individuals face substantial barriers to accessing evidence-based interventions. Advances in generative artificial intelligence (AI), particularly large language models (LLMs), have generated optimism about improving access and care. We present five emerging use cases for clinical AI tools in the context of PTSD treatment, some of which were presented as part of a symposium at the 40th Annual Meeting of the International Society for Traumatic Stress Studies. The first two use cases involve AI-assisted training tools. The third use case focuses on an AI-assisted automated fidelity rating system aimed at improving adherence to evidence-based PTSD protocols. The last two use cases feature AI-assisted therapy tools. Although AI-based innovations hold the promise of enhancing the reach and consistency of evidence-based PTSD interventions, they also raise important ethical and safety challenges, including risk of bias, threats to patient privacy, and the question of how to incorporate clinical oversight. Ongoing collaboration among multidisciplinary teams involving clinicians, researchers, and technology developers will be essential to ensure that AI tools remain patient-centered, ethically sound, and effective.

Generative Artificial Intelligence in PTSD Treatment: Exploring Five Different Use Cases

2025-05-15

Posttraumatic stress disorder (PTSD) is a prevalent and debilitating condition, yet many individuals face substantial barriers to accessing evidence-based interventions. Advances in generative artificial intelligence (AI), particularly large language models (LLMs), have generated optimism about improving access and care. We present five emerging use cases for clinical AI tools in the context of PTSD treatment, some of which were presented as part of a symposium at the 40th Annual Meeting of the International Society for Traumatic Stress Studies. The first two use cases involve AI-assisted training tools. The third use case focuses on an AI-assisted automated fidelity rating system aimed at improving adherence to evidence-based PTSD protocols. The last two use cases feature AI-assisted therapy tools. Although AI-based innovations hold the promise of enhancing the reach and consistency of evidence-based PTSD interventions, they also raise important ethical and safety challenges, including risk of bias, threats to patient privacy, and the question of how to incorporate clinical oversight. Ongoing collaboration among multidisciplinary teams involving clinicians, researchers, and technology developers will be essential to ensure that AI tools remain patient-centered, ethically sound, and effective.

Evaluating effectiveness and engagement strategies for asynchronous, messaging, trauma-focused therapy for posttraumatic stress disorder: Study design and methodology for a hybrid effectiveness-implementation randomized controlled trial

Contemporary Clinical Trials · 2025-11-20

TherapyTrainer: Using AI to train therapists in written exposure therapy

2025-06-25

Though evidence-based treatments for mental disorders are effective, existing implementation efforts are expensive and difficult to scale. Novel solutions— especially those that offer active learning strategies, repeat skill practice and personalized feedback to therapists — are needed to fill this gap. We developed TherapyTrainer, which uses large language models (LLMs) to allow therapists to practice delivering written exposure therapy (WET) for PTSD to AI-Patients while receiving expert feedback from an AI-Consultant. Here we present initial feasibility, acceptability and usability data for TherapyTrainer gathered from therapists, supervisors, and WET expert-consultants across iterative rounds of development. In Phase 1, we rapidly prototyped and developed TherapyTrainer based on ongoing feedback from WET clinicians and experts (n = 4). In Phase 2, mixed methods data from therapists engaged in an otherwise-routine WET workshop (n = 14) indicated that TherapyTrainer is feasible and acceptable and may help therapists feel prepared to deliver WET. In Phase 3, therapists (n = 6) completed structured user testing interviews to identify key issues impacting usability for subsequent rounds of development. AI and large language models hold potential to provide ongoing support to therapists in a cost-effective and scalable manner, and may help close the research-practice gap.