JMIR Mental Health

Governing AI for Mental Health: Fragmented State Approaches and the Case for a Federal Framework

2026-07-28T07:00:20-04:00

State-level regulation of AI used for mental health is emerging in the absence of a federal framework. States are taking different approaches to regulation, resulting in a fragmented regulatory landscape. This Viewpoint aims to identify the governance approaches that US states are using to regulate the use of AI in mental health and analyze the limitations of each. A 4-state case analysis was conducted using the statutory text of bills and laws in Illinois, Utah, New York, and Nevada. Two governance approaches were identified. The first regulates the use of AI in clinical contexts, and the second regulates the technology itself. Some states have combined elements of both approaches to address AI use more comprehensively. While these approaches aim to mitigate harm, they differ in where they believe risk lies in the use of AI for mental health support. The limitations of these divergent approaches include uneven protections for consumers and regulatory uncertainty for developers, vendors, deployers, and clinicians. Because AI in mental health operates across both clinical and consumer domains, neither approach alone can address the risks associated with its use for mental health support. A coordinated, risk-based federal regulatory floor is needed to ensure consistent protections across states.

Structuring Digital Mental Health Care Navigation: Co-Design Nominal Group Technique Study to Develop the MChart Definition and Typology of the Characteristics of Digital Mental Health Care Navigation Tools

2026-07-27T16:15:18-04:00

Background: Australia’s mental health care system has been characterized by complexity and fragmentation, as highlighted by numerous reports, commissions, and inquiries. In response, digital mental health care navigation tools have emerged as a promising solution to help individuals locate appropriate mental health services. The rapid proliferation of these tools—without a clear understanding of their definitions and characteristics—risks creating confusion rather than clarity for users. Terms such as “navigation” and “navigators” are often used interchangeably, further complicating the landscape. Objective: This study addressed the need for a standardized definition and typology of the characteristics of digital mental health care navigation tools. Methods: This study was part of the development of a digital mental health care navigation tool for navigators and planners (MChart). It used a co-design approach using expert-based cooperative analysis, which is a nominal group technique to develop a definition and typology of the characteristics of digital mental health care navigation tools. This process was guided by the Technology Readiness Level for Implementation Sciences framework. The co-design process involved two 2-hour sessions with an expert panel comprising 28 participants, including representatives from mental health planning, primary health care, health care financing and delivery, community-managed organizations, clinical settings (psychiatrists, psychologists, and general practitioners), and consumers. Results: The expert panel collaboratively developed a consensus definition of digital mental health care navigation tools, outlining their scope and intended targets. Through the co-design process, the panel identified 157 characteristics of digital mental health care navigation tools. These characteristics were organized into 5 primary domains: type, management, content, design, and quality. The definition and typology characteristics provide a structured framework for understanding and evaluating the diverse range of digital mental health care navigation tools currently available. Conclusions: The co-designed definition and typology offer a foundational step toward reducing confusion in the digital mental health care navigation space. This study supports the development of quality standards that can be used to assess and compare existing and future tools. This framework has the potential to guide developers, end users, and policymakers in creating more effective, user-centered navigation solutions within Australia’s mental health care system and internationally.

Clinical Outcomes and Predictors of Improvement With Virtual Behavioral Health Care for Gambling Disorder: Retrospective Cohort Study

2026-07-24T15:01:09-04:00

Background: Gambling disorder is associated with substantial psychiatric and functional burden, yet few individuals receive treatment. Limited real-world evidence exists evaluating outcomes of virtually delivered behavioral health care for gambling disorder, particularly among clinically complex patients. Objective: The purpose of this study was to evaluate gambling symptom severity outcomes among adults with gambling disorder receiving care from Birches Health. We aimed to (1) characterize the clinical profile of adults seeking treatment for gambling disorder; (2) quantify changes in gambling symptom severity over the initial 12 weeks of treatment and examine whether baseline clinical complexity, such as gambling symptom severity, depression severity, and psychiatric comorbidities, was associated with differences in gambling symptom severity improvement over time; and (3) estimate the timing and likelihood of achieving clinically meaningful improvement in gambling symptom severity. Methods: This retrospective cohort study included 1305 adults receiving virtual behavioral health treatment for gambling disorder through Birches Health between June 2024 and April 2026. Gambling symptom severity was assessed using the Gambling Symptom Assessment Scale (G-SAS) weekly. Linear mixed-effects models evaluated changes in gambling symptom severity over 12 weeks and associations with baseline clinical characteristics. Clinically meaningful improvement was defined as a reduction of 4 or more points in the G-SAS score. Results: Participants had a mean age of 41.5 (SD 13.1) years, 65.2% (851/1305) were male, and baseline gambling symptom severity was moderate (mean G-SAS score 20.5, SD 11.83). Over half (730/1305, 56%) of participants presented with at least one psychiatric comorbidity, most commonly anxiety disorder (351/1305, 26.9%) and depressive disorder (276/1305, 21.1%). Gambling symptom severity declined significantly over the first 12 weeks of treatment, with G-SAS scores decreasing by approximately 0.099 points per day (P<.001), corresponding to an estimated 8.3-point reduction over 12 weeks. Higher baseline depressive symptom severity was associated with faster improvement in gambling symptoms (P=.01), whereas depressive disorder (P=.03) and attention-deficit/hyperactivity disorder (P=.008) diagnoses were associated with slower improvement trajectories. Among patients with routine follow-up assessments recorded during the initial 12 weeks of treatment (1071/1305, 82.1%), 71.7% (935/1305) achieved clinically meaningful improvement in gambling symptom severity, with a median time to improvement of 14 days. Conclusions: A clinically complex population of adults receiving care through a national virtual behavioral health care provider demonstrated rapid and clinically meaningful reductions in gambling symptom severity. These findings highlight the potential of specialized virtual care models to expand access to gambling treatment and support symptom improvement in routine care settings. Future research should evaluate longer-term recovery trajectories and identify factors associated with sustained improvement and ongoing engagement in care.

A Digital Acceptance and Commitment Therapy and Education Intervention for Caregivers of Very Preterm Infants in the Neonatal Intensive Care Unit: Randomized Controlled Trial

2026-07-23T16:45:12-04:00

Background: Parents of very preterm infants admitted to the neonatal intensive care unit (NICU) experience high levels of psychological distress, yet access to timely, evidence-based mental health support is limited by staffing and resource constraints. Digital mental health interventions offer a scalable approach to addressing this gap; however, their effectiveness has not been well established in NICU caregiver populations, particularly during periods of acute stress. Objective: This study aims to evaluate the effectiveness of a self-guided digital acceptance and commitment therapy (ACT)–based intervention combined with NICU-specific education (NICU parent acceptance and commitment therapy [NPACT]). The study explored the intervention’s effects on stress among parents and primary caregivers of very preterm infants, compared to a digital education-only intervention, and active control. Methods: We conducted a 3-arm, single-center, randomized controlled cluster trial in a tertiary NICU. Parents and primary caregivers of very preterm infants (<32 wk’ gestational age,<1 wk old) were randomized by family cluster to (1) NPACT (ACT+ education), (2) a digital education-only intervention, or (3) active control. Digital interventions were delivered via a web-based platform over 2 weeks. The primary outcome was NICU-related stress on the Parent Stressor Scale: Neonatal Intensive Care Unit (PSS:NICU) at 2 weeks postrandomization. Secondary outcomes included caregiver anxiety, depression, perceived stress, and selected neonatal outcomes. Engagement and perceived helpfulness were assessed for digital interventions. Results: A total of 102 caregivers from 68 family clusters (79 infants; mean gestational age 28.1, SD 2.2 wk) were enrolled. There were no statistically significant between-group differences in the mean PSS:NICU scores at 2 weeks (NPACT 3.0, SD 0.9; education-only 2.5, SD 1; active control 2.6, SD 0.9; adjusted mean difference for NPACT vs active control 0.04, 95% CI −0.39 to 0.47). No between-group differences were observed for secondary psychological outcomes at any time point. However, caregivers in both digital intervention groups had higher odds of full breastfeeding at discharge compared with active control. Engagement with the digital interventions was high, with 97% (28/29) of NPACT participants and 76% (19/25) of education-only participants completing at least 5 of 7 modules, and both interventions were rated as very helpful. Conclusions: In this trial, an unguided digital mental health intervention delivered during NICU admission did not reduce NICU-specific parental stress or other psychological outcomes relative to active control. However, the intervention was highly used by caregivers. These findings suggest that while a brief digital mental health intervention can be successfully implemented in a high-stress clinical setting with caregivers, its capacity to reduce acute psychological distress may be limited. Secondary findings indicate potential benefits of the digital intervention on breastfeeding, generating hypotheses for future research. Digital mental health interventions in neonatal settings may be most effective when integrated within hybrid models of care and/or delivered beyond the acute admission phase. Trial Registration: Australian New Zealand Clinical Trials Registry ACTRN12623000641695; https://tinyurl.com/2e8677bb International Registered Report Identifier (IRRID): RR2-10.1016/j.cct.2024.107519

Optimizing Treatment Strategies in the Bipolar Disorder Spectrum With Classical AI Approaches: Systematic Review of Performance, Bias, and Clinical Applicability

2026-07-21T14:00:20-04:00

Background: Bipolar disorder (BD) is a complex and heterogeneous psychiatric condition, characterized by fluctuating clinical courses that affect approximately 1%‐2% of the global population in their lifetime. Despite pharmacological advances, treatment response varies significantly among patients, making the identification of individualized treatment strategies a major challenge. Artificial Intelligence (AI), through its classical approaches, has emerged as a powerful tool in precision psychiatry to identify subtle patterns in complex data and inform personalized clinical decisions. Objective: The present systematic review aimed to examine the current evidence on classical AI-supported treatment optimization in the BD spectrum. Methods: The review was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines. Four databases (PubMed, Web of Science, Scopus, and Embase) were searched for original studies published after 2015 on the application of classical AI in the treatment of BD in adult patients. Publication bias was evaluated by visual inspection of a funnel plot. The methodological quality, risk of bias, and clinical applicability of the predictive models were assessed using the Prediction Model Risk Of Bias Assessment Tool for prediction models using regression or AI methods (PROBAST+AI; PROBAST+AI Working Group) tool. Results: A total of 35 studies were included and classified into 5 outcome-based categories, including acute symptomatic response, long-term maintenance response, relapse and readmission risk, safety and dose optimization, and brain aging and phenotyping. Acute symptomatic response models performed modestly (pooled area under the curve [AUC] 0.68), while imaging improved accuracy (74%‐77%). Long-term maintenance response models showed moderate-to-high performance (pooled AUC 0.80), with biomarker- and cellular-based models reaching 96%‐99% accuracy. Relapse and readmission prediction achieved a pooled AUC of 0.71, with digital phenotyping and rule-based methods performing best (AUC 0.85‐0.88). Safety and dose optimization models achieved 85%‐97% accuracy. Brain aging and phenotyping studies highlighted accelerated brain aging in BD, partially mitigated by lithium, and revealed novel data-driven subgroups. However, 3 studies were considered at high risk of bias due to small sample sizes associated with disproportionately high-performance estimates. An additional study was identified as potentially biased because it lay markedly distant from the funnel plot’s confidence line. Finally, the PROBAST+AI assessment revealed a high risk of bias in most studies, primarily due to data analysis limitations, small sample sizes, and lack of external validation. Conclusions: The adoption of classical AI tools in BD serves as a driver for therapeutic optimization, although current AI tools in BD should still be considered exploratory rather than ready for clinical use. Effective implementation in real-world clinical scenarios requires more robust, transparent, and externally validated models to ensure reliability and generalizability.

Immersive Technologies in Forensic Mental Health and Prison Settings: Scoping Review

2026-07-17T17:45:02-04:00

Background: The application of immersive technologies, particularly virtual reality, has expanded rapidly across health care domains, including mental health, rehabilitation, and education. These technologies enable the creation of controlled, interactive, and ecologically valid environments that can support therapeutic interventions, skill development, and behavioral assessment. Within forensic mental health services (FMHS) and prison settings, where individuals often present with complex psychological needs in restrictive and highly regulated environments, immersive technologies offer potential advantages such as safe simulation of real-world scenarios, enhanced engagement, and personalized intervention delivery. However, despite increasing interest, the evidence base remains fragmented, and questions persist regarding effectiveness, ethical implications, and feasibility of implementation in secure and resource-constrained contexts. Objective: Interest in immersive technologies in FMHS and prison settings is growing, yet their role remains unclear. This scoping review mapped current uses, highlighted opportunities, and identified key gaps and considerations for future implementation. Methods: A scoping review of English-language publications (2010-2025) was conducted using the Scopus, PubMed, and CINAHL databases. Data extraction followed the Joanna Briggs Institute framework, and thematic analysis explored benefits, drawbacks, and implementation barriers. Results: Thirty sources were identified. Primary research focused mainly on virtual reality for therapy, skill training, education, and assessment. There was evidence suggesting benefits such as increased engagement, emotional regulation, skill acquisition, autonomy, and improved clinician-patient dialogue. However, the studies were small, heterogeneous, and inconsistently reported, with limited long-term follow-up. Implementation barriers included institutional, ethical, and technical constraints and limited personalization and end user involvement. Co-design and participatory approaches surfaced as key enablers of acceptability, relevance, and safe use. Conclusions: The existing evidence base is preliminary and exploratory but indicates that immersive technologies may have potential value in FMHS and prison contexts. Current findings should be interpreted cautiously because studies are small, heterogeneous, and rarely include long-term follow-up. More robust evidence, careful implementation, and meaningful end user input are needed to support safe, relevant, ethical, and effective use. The emphasis on coproduction and guidance for safe, user-centered implementation is a novel contribution.

Temporal Patterns of Engagement and Sentiment in a Suicide Prevention Mobile App: Three-Year Observational Study

2026-07-16T17:30:14-04:00

Background: Temporal fluctuations in distress and suicidal ideation across daily, weekly, and seasonal cycles may influence the use and effectiveness of digital suicide prevention tools. Understanding patterns of app engagement, perceived suffering, and affective expression can inform the design of proactive, personalized digital interventions, thereby impacting adherence and efficacy. Objective: This study aimed to examine temporal patterns of engagement with 2 core components of the suicide prevention app SERO (Suicide Prevention: a Uniform Effort, Resource-Oriented; BFH, Lucerne Psychiatry), specifically the safety plan and the PRISM-S (Pictorial Representation of Illness and Self-Measure—Suicidality) self-assessment, using 3 years of interaction log data, assessing variations across circadian, weekly, and seasonal cycles, and evaluating the sentiment of free-text responses submitted immediately after PRISM-S self-assessments. Methods: We analyzed anonymized interaction logs from the SERO app collected over 3 years (November 2022 to December 2025). Engagement metrics included the frequency of use of the safety planning functionality and PRISM-S self-assessment entries. Free-text responses provided after PRISM-S assessments were analyzed using automated sentiment classification. Temporal analyses examined variations by the hour of the day, day of the week, and season. One-way ANOVAs, post hoc tests, and Pearson correlations were used to examine patterns and associations between perceived suffering and sentiment. Results: A total of 1076 users engaged with the safety planning functionality of the SERO app, generating 3502 entries, with coping strategies and warning signs showing the highest mean interactions and personal beliefs the lowest. Separately, 1212 app users accessed the PRISM-S self-assessment, producing 2329 entries (mean distance 12.91, 95% CI 12.39‐13.42 cm), with most app users recording only 1 or 2 registrations. Safety planning engagement showed clear diurnal patterns, peaking in the afternoon (2 PM to 3 PM) and being lowest at night (midnight to 3 AM), whereas PRISM-S scores were stable across time. Sentiment analysis revealed predominantly negative affect (mean score of −0.41, SD 0.51, 95% CI −0.44 to −0.39), correlated with PRISM-S distance, and was most negative at night (specifically at 11 PM) and during the afternoon (2 PM to 5 PM). Seasonal effects were small but significant for PRISM-S, with the lowest perceived suffering in summer. Conclusions: Digital suicide prevention tools can support routine patterns of coping behavior, but periods of increased reported distress, particularly at night, may be underaddressed. Integrating automated sentiment analysis alongside self-assessments could potentially enable personalized, time-adaptive interventions that detect changes in emotional state and deliver timely, tailored support, thereby strengthening proactive engagement and resilience.

A Self-Guided Mobile Mindfulness Intervention Embedded in Daily Routines for Adults With Mild to Moderate Psychological Distress: Randomized Controlled Trial

2026-07-15T17:15:08-04:00

Background: Mobile mindfulness interventions have shown promise for reducing anxiety and depressive symptoms, but sustaining engagement remains a persistent challenge. Many digital programs still rely on formal practice that requires dedicated time, which may be difficult to integrate into daily life. Objective: This randomized controlled trial evaluated Habitual Mindfulness Practice (HMP), a self-guided mobile mindfulness intervention that embeds brief practices into recurring daily routines, among adults with mild to moderate psychological distress. Outcomes were compared with those of Traditional Mindfulness (TM), Mindfulness-Based Psychoeducation (MBP), and a waitlist control (WL). Methods: Adults aged 18 to 65 years with mild to moderate symptoms of anxiety or depression were randomly assigned in a 1:1:1:1 ratio to HMP, TM, MBP, or WL (N=686). All procedures were conducted online, and the intervention was fully self-guided, with outcomes assessed using self-report measures. The intervention lasted 21 days, with assessments conducted at baseline, postintervention, and 3-month follow-up. Primary outcomes were depressive and anxiety symptoms. Secondary outcomes included mindfulness, cognitive emotion regulation, affective balance, and interpersonal difficulties. Postintervention group differences were examined, controlling for baseline scores, and longitudinal trajectories were evaluated across the active intervention conditions. Results: At postintervention, significant group effects were observed for depressive symptoms (=28.67, <.001, ηp²=0.11) and anxiety symptoms (=30.11, <.001, ηp²=0.12). Both HMP and TM showed lower depressive and anxiety symptom scores than MBP and WL. TM showed lower postintervention anxiety than HMP (=.04), whereas depressive symptoms did not differ significantly between HMP and TM (=.63). The mindfulness practice conditions also showed more favorable postintervention outcomes for mindfulness, affective balance, interpersonal difficulties, and emotion regulation. Improvements in depressive and anxiety symptoms were generally maintained at follow-up among the active intervention conditions, although maintenance of secondary outcomes varied across measures. Postintervention outcome data were available for 55.2% (379/686) of randomized participants, and follow-up outcome data were available for 24.1% (124/515) of participants in the active intervention conditions. HMP and TM did not differ significantly in practice duration, engagement, or satisfaction. Conclusions: A routine-embedded, self-guided mobile mindfulness intervention may be a feasible approach for reducing mild to moderate psychological distress. HMP produced benefits broadly comparable to those of traditional app-delivered mindfulness, but it did not confer advantages in engagement or short-term efficacy. Trial Registration: Chinese Clinical Trial Registry ChiCTR2400093771; https://tinyurl.com/3d6tky8v

Active Ingredients in Digital Cognitive Interventions: Integrating Dismantling Designs With Mechanistic Neuroscience

2026-07-14T16:15:11-04:00

Digital cognitive interventions (DCIs) have emerged as scalable approaches for treating cognitive dysfunction across psychiatric, neurological, and aging populations. Despite growing evidence of efficacy, little is known about which intervention components drive therapeutic effects or through which neurocognitive mechanisms they operate. As a result, null findings are often difficult to interpret, making it unclear whether interventions failed to engage their intended targets, or whether the targets themselves are not causally related to meaningful outcomes. This limits intervention refinement, comparative evaluation, and precision personalization. Here, we argue that DCI research should shift from broad efficacy testing toward mechanistic trials designed to identify active ingredients—the intervention components responsible for engaging prespecified neurocognitive targets and producing clinically meaningful benefits. We propose adapting dismantling design methodology from psychotherapy research in order to integrate Research Domain Criteria constructs, mechanistic neuroscience, and high-resolution digital behavioral data to identify factors driving cognitive and functional outcomes. This approach aligns with the National Institute of Mental Health experimental therapeutics framework by explicitly linking target specification and target engagement with downstream clinical and functional outcomes. Mechanistic dismantling trials can determine whether specific DCI features, including adaptive difficulty, reward schedules, feedback contingencies, task variability, cognitive targets, and human support, are necessary, sufficient, or synergistic for engaging neural circuitry and producing durable and clinically meaningful transfer. Beyond optimizing intervention design, such studies may transform null or negative trials into mechanistically interpretable findings, while clarifying disease mechanisms and supporting the development of personalized, optimized, and usable DCIs.

AI Agents Are Coming: 5-Stage Taxonomy of Language-Based AI Systems for Psychiatry, Psychotherapy, and Counseling

2026-07-13T11:30:03-04:00

The rapid evolution of large language models has accelerated the development of agentic artificial intelligence (AI) systems capable of pursuing autonomous goals, creating an urgent need for structural frameworks in psychiatry and psychotherapy. While existing classifications often draw parallels to autonomous driving, this paper argues that the mental health domain requires a distinct, domain-specific theoretical foundation, as the 2 domains differ fundamentally in their semantic, ideographic, and epistemological demands. Furthermore, they differ in their end goals, for which we introduce terms such as agentic guidance capability. To guide clinicians and researchers through these developments, we propose a 5-stage taxonomy for language-based AI systems that differentiates technical functionality from clinical effectiveness. The taxonomy progresses from level 1 (knowledge level), in which systems perform static benchmark tasks, to level 2 (elementary level), characterized by dynamic engagement in specific therapeutic microskills. At level 3 (integration level), systems achieve consistency across and within modules, as well as basic case-level conceptualization suitable for blended therapy under human oversight. Level 4 (saturation level) describes therapist-in-the-loop systems capable of autonomous functioning with minimal supervision, whereas level 5 (mastery level) represents AI systems that are technically capable of performing autonomous therapy. By distinguishing technical functionality from clinical effectiveness, we conclude that level 4 or level 5 performance does not automatically translate into full treatment effectiveness, even if high treatment fidelity can be achieved. We conclude by emphasizing the need to shift benchmarking from static knowledge tests to dynamic evaluations of therapeutic capabilities in order to safely navigate the transition toward autonomous care.