Background Timely follow-up is crucial for youth whose first mental health contact is through a hospital; virtual care could facilitate this. Objective This study aimed to examine the association between the significant expansion of virtual care in Ontario, Canada, and follow-up mental health care rates for these youths. Methods We conducted a population-based repeated cross-sectional study using linked health administrative databases. We identified first-contact emergency department (ED) visits or hospital admissions for self-harm, mood disorders, or psychosis from March 1, 2009, to February 29, 2024, among Ontario youth (aged 10-24 years). Virtual care was expanded in March 2020; we examined 2 expansion stages (temporary and permanent). Autoregressive integrated moving average analyses estimated level and slope changes (vs before virtual care) in age- and sex-standardized monthly rates of 7-day mental health follow-up. Subgroup analyses examined rurality and socioeconomic status. Results We identified 42,208 first-contact ED visits and 17,701 first-contact hospital admissions. Pre–virtual care mean follow-up rates were 17.64 (SD 2.97) per 100 first-contact ED visits and 21.90 (SD 5.05) per 100 first-contact hospital admissions. Temporary expansion was associated with immediate increases in follow-up that declined (level changes: 4.81, 95% CI 2.82-6.81 for ED visits and 7.51, 95% CI 4.41-10.60 for hospital admissions; slope changes: −0.31, 95% CI −0.42 to −0.21 for ED visits and −0.32, 95% CI −0.48 to −0.15 for hospital admissions). Permanent expansion was associated only with an immediate decrease in ED visit follow-up (level change: −3.81, 95% CI −6.70 to −0.92) that was significant in rural (−1.06, 95% CI −1.93 to −0.184) but not urban (0.44, 95% CI −0.179 to 1.06) areas; results were otherwise similar between subgroups. Conclusions Virtual care was not associated with sustained changes or equity improvements in mental health follow-up for high-acuity youth, representing a missed opportunity to extend physician reach and address barriers. Further research and quality improvement are needed.
Background:Major depressive disorder affects over 280 million people worldwide, and access to effective treatment remains limited. Transcranial direct current stimulation (tDCS) is a noninvasive option, and portable devices now allow for home-based delivery under varying degrees of remote supervision. Objective:This study aimed to systematically review and meta-analyze the efficacy, safety, feasibility, and acceptability of home-based and remotely supervised tDCS for depressive disorders. Methods:Following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 and PRISMA-S (Preferred Reporting Items for Systematic Reviews and Meta-Analyses literature search extension) guidelines, we searched MEDLINE, Embase, Web of Science, the Cochrane databases, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform up to July 2025, with backward and forward citation searching. Two reviewers independently screened records, extracted data, and assessed risk of bias (version 2 of the Cochrane risk-of-bias tool for randomized trials, Newcastle-Ottawa Scale for observational studies, and Critical Appraisal Skills Programme for qualitative studies) and certainty of evidence (Grading of Recommendations Assessment, Development, and Evaluation; GRADE). Results:This review included 12 distinct studies (16 reports), of which 6 (50%) were randomized sham-controlled trials forming the meta-analytic pool. Active home-based tDCS produced a small, statistically significant improvement over sham (pooled Hedges g=0.36, 95% CI 0.06-0.66; P=.03; I2=34.3%). The effect was not robust to removal of the single largest positive trial (omitting the one study from 2025: g=0.39, 95% CI -0.12 to 0.91), and trial-level results were mixed: the 2 largest trials (one unsupervised [n=210] and one self-administered [n=141]) were negative on their primary depression outcomes, whereas the largest real-time supervised trial (n=174) was positive (between-group 95% CI 0.51-4.01; P=.01). This estimate was concordant in direction with an independent peer-reviewed meta-analysis of overlapping trials, which reported a pooled Montgomery-Åsberg Depression Rating Scale reduction (weighted mean difference -2.74, 95% CI -4.19 to -1.29) and Hamilton Depression Rating Scale reduction (weighted mean difference -2.24, 95% CI -4.16 to -1.49), attenuating to nonsignificance (P>.05) in major depressive disorder without comorbid cognitive impairment. The pooled effect fell at or near the minimal clinically important difference. GRADE certainty was moderate. Adverse events were predominantly mild: one pilot study was terminated early for skin lesions, and one nonfatal suicide attempt occurred in an unsupervised trial. Conclusions:Home-based and remotely supervised tDCS produces a small, statistically significant but clinically modest antidepressant effect that is sensitive to the inclusion of the largest positive trial, with the 2 largest trials being negative. The available controlled evidence does not establish supervision intensity as a determinant of efficacy. Current data are insufficient to recommend routine clinical adoption; adequately powered trials with standardized supervision and longer follow-up are needed.
Abstract BackgroundMental health chatbots are increasingly used to support people with depressive symptoms, and large language models make these systems more flexible than rule-based chatbots. However, it remains unclear how well large language model–based chatbots deliver structured psychological interventions. ObjectiveThis study examined how well a GPT-4o–based chatbot delivered a behavioral activation intervention for young people with depression using sessions with artificial users and clinical expert assessment. It also identified limitations and potential refinements. MethodsWe implemented a GPT-4o (gpt-4o-2024-08-06; OpenAI)–based chatbot using a structured system prompt to deliver a single-session behavioral activation intervention for people with depression aged 14 to 29 years. We generated 48 sessions with GPT-4o–based artificial users derived from clinical vignettes varying across 7 characteristics. Ten clinical experts, either licensed psychotherapists or advanced psychotherapy trainees, independently assessed the sessions using the 14-item Quality of Behavioral Activation Scale (Q-BAS), rated from 0 to 6, supplemented by rating therapeutic capabilities, artificial user authenticity and difficulty, and qualitative feedback. ResultsThe chatbot completed all 7 intervention phases in every session. The mean holistic session quality rating was 3.94 (SD 1.23), and the mean Q-BAS rating was 4.03 (SD 1.18). Thirteen of 14 Q-BAS components exceeded the satisfactory threshold of 3. Ratings were highest for mood assessment (mean 5.42, SD 1.09) and activity planning (mean 4.98, SD 1.41) and lowest for explaining positive reinforcement (mean 2.92, SD 2.30) and supporting activity-mood monitoring (mean 3.02, SD 2.04). Therapeutic capability ratings were highest for message safety (mean 5.90, SD 0.37), message clarity (mean 5.56, SD 0.77), and objective, nonjudgmental communication (mean 5.17, SD 1.04) and lowest for therapeutic rapport (mean 4.12, SD 1.45) and natural conversation flow (mean 4.25, SD 1.42). Artificial users were rated below the scale midpoint for authenticity (mean 2.75, SD 1.41) and difficulty (mean 1.23, SD 1.46). Clinical experts described the chatbot as structured, clear, and safe but identified insufficient clinical reasoning as the main limitation, particularly in evaluating the therapeutic suitability and feasibility of activities, barriers, solution strategies, and rewards. Artificial users were often highly compliant, especially when identifying positive activities. ConclusionsIn expert-rated sessions with artificial users, the chatbot delivered the behavioral activation intervention as intended and performed strongest on procedural components. It performed less well on positive reinforcement and activity-mood monitoring, indicating refinement needs in clinical reasoning, follow-up questioning, and evaluating whether proposed activities, plans, barriers, solution strategies, and rewards are therapeutically appropriate and feasible. The findings identify targets for improvement before testing with human users, while the artificial user design and expert ratings limit conclusions about real therapeutic interactions.
Background:Suicide remains a leading cause of death among young adults aged 18 to 25 years. Young adults experiencing suicidal ideation (SI) are increasingly using crisis text services (CTSs), a free and accessible option for crisis intervention. Little is known about CTSs from the young adult perspective. Objective:This study aimed to characterize young adults' experiences with and perceptions of CTSs for SI. Methods:We conducted in-depth interviews, by phone, Zoom, or text, with young adults (n=39) in the United States who had a lifetime history of SI. Participants included those who had or had not engaged with CTSs for SI. Semistructured interviews were conducted from January to July 2024. The data were analyzed using a modified grounded theory approach. Results:We constructed 5 key themes to characterize young adults' perceptions of and experiences with CTSs for SI. Young adults perceived CTSs as a unique component of their mental health crisis management. They appreciated CTSs' technological features, particularly the privacy they provided and the ability to reflect on and edit responses. However, they expressed dissatisfaction with the nonspecific nature of many CTS interactions. The perceived anonymity of CTSs served multiple functions, both as a motivator for CTS use and as a potential point of vulnerability, should it be lost during a CTS interaction. Participants' perceptions of CTSs' impact varied; some viewed them as beneficial, whereas others reported neutral or inconsistent effects over time. Conclusions:Among young adults with SI, CTSs are a key yet imperfect resource. Quality improvement and evaluation efforts may be needed to understand how responders can better tailor responses to improve conversational quality and consistency for young adult texters.
Large language models are increasingly used within and alongside therapy. As large language models perform more therapy functions, questions arise about what the future might hold for therapists and what they will do. We argue that the enduring therapist role in the age of AI-assisted care is currently best understood through relational, adaptive, and accountability functions. These functions include therapeutic challenge, use of the therapeutic relationship as a mechanism of change, rupture detection and repair, bearing witness to suffering, calibration of pace and treatment burden, and clinical judgment under uncertainty across the broader care pathway. Drawing on psychotherapy theory, digital mental health research, the declarative-procedural-reflective model by Bennett-Levy, and our clinical experience, we propose a clinically informed, hypothesis-generating, relational-adaptive-accountability framework. This framework is intended to support further empirical testing and may have implications for workforce development, supervision, training, and service design.
