Mental health disorders affect 970 million people globally; yet, over 50% do not access timely evaluation due to structural barriers and professional shortages. Chatbots and AI-based conversational agents have emerged as promising tools for mental health screening and assessment.

This study systematically evaluated the effectiveness, accuracy, reliability, and acceptability of chatbots and AI-based conversational agents for mental health screening and assessment in adults.

Systematic search conducted in May 2025 across PubMed/MEDLINE, PsycINFO, Scopus, and Web of Science (2019-2025), following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines. Eligible studies evaluated chatbots or AI for mental health screening/assessment in adults (≥18 y). Risk of bias was assessed using appropriate tools (Risk of Bias 2, Quality Assessment of Diagnostic Accuracy Studies-2, Joanna Briggs Institute checklists, and Mixed Methods Appraisal Tool). This systematic review was registered with PROSPERO (International Prospective Register of Systematic Reviews; CRD420251072392).

Eighteen studies (2021-2025) were included, with samples ranging from 20 to 3902 participants. Rule-based chatbots demonstrated high reliability (Cronbach α >0.85) and good acceptability (Acceptability of Intervention Measure >19/25). Generative models (large language models) achieved sensitivities of 0.84 to 0.93 and specificities of 0.80 to 0.96 for depression and anxiety, with correlations up to r=0.96 with expert clinicians in suicide risk assessment. Hybrid approaches combining large language models with machine learning achieved exceptional performance for cognitive impairment (F₁-score of 92.1%, specificity of 99.6%). Most studies reported high user satisfaction (≥70%), although barriers existed among older populations. Methodological quality was heterogeneous with a moderate risk of bias in critical dimensions.

Chatbots and AI conversational agents demonstrate clinically relevant performance in mental health screening and assessment. However, safe implementation requires clear clinical protocols, professional supervision, integration with electronic health records, and active mitigation of algorithmic bias. These technologies should complement rather than replace clinical judgment.

Mental Health Online, MindSpot, and THIS WAY UP are 3 Australian Government-funded digital mental health services (DMHSs). These services are free for consumers and deliver a range of (depression and anxiety disorder) psychological assessments and interventions using the internet, with or without clinician support.

This study examines the uptake and effectiveness of these services.

We used 3 data sources to examine the performance of the three DMHSs: (1) aggregated routinely collected service data to describe uptake of the services from January 2013 to December 2021, (2) peer-reviewed and gray literature reporting treatment outcomes, and (3) treatment outcome data from one evaluation conducted by Mental Health Online. The second and third data sources were used to analyze the effectiveness of the DMHSs compared to other mental health treatment programs (Australian primary, public, and low-intensity mental health care; UK stepped psychological care). We generated pooled mental health symptom (Cohen d) effect sizes for each service's clinician-supported and self-directed treatments and compared those with effect sizes we calculated for the alternative treatment programs.

The 3 DMHSs offered care to more than 282,000 consumers and therefore contributed to improving overall access to mental health care in Australia. Clinician-supported online treatment significantly improved the mental health of consumers who use these services (Mental Health Online, Cohen d=0.95; MindSpot, Cohen d=1.42; and THIS WAY UP, Cohen d=1.04), and self-directed treatment by Mental Health Online produced a moderate reduction (Cohen d=0.59) in anxiety disorder severity ratings. Clinician-supported treatments produced improvements in mental health symptoms that are close or equivalent to most (face-to-face, phone, and stepped or symptom severity-matched mental health care) comparison treatments examined.

Mental Health Online, MindSpot, and THIS WAY UP are producing clinically significant improvement for consumers experiencing psychological distress and anxiety and depression symptoms. The magnitude of improvement produced by clinician-supported treatment is comparable with more resource-intensive face-to-face treatment options. Heterogeneity in the magnitude of improvements produced by the 3 DMHSs (and comparator interventions across levels of clinical need) indicates that findings should be interpreted with caution, particularly in relation to the populations for whom DMHSs may be most appropriate.

Mental health is a central concern for aging and well-being over the life course. This study examined ethno-racial patterns of depressive symptoms and loneliness among a diverse cohort of U.S. midlife adults to illuminate why some individuals age more successfully than others. Drawing on the Stress Process Model, we investigated whether these patterns persist after accounting for socioeconomic and relational stressors and social support resources.

We used new Wave VI data from the National Longitudinal Study of Adolescent to Adult Health (Add Health). Descriptive statistics and negative binomial regression were employed to document ethno-racial variation in depressive symptoms and loneliness among this cohort of midlife U.S. adults (N = 11,435; ages 39-50).

Multiracial adults initially reported slightly higher depressive symptoms and loneliness relative to White adults, but these differences were attenuated with adjustment for stressors and social support. Black adults exhibited poorer mental health than White adults descriptively, but better outcomes after adjusting for stressors and social support. Hispanic adults reported lower levels of depressive symptoms and loneliness than White adults net of sociodemographic covariates. Asian adults demonstrated no statistically significant differences from White adults.

Ethno-racial groups each display unique mental health patterns in midlife, possibly reflecting differences in how ethno-racial groups perceive and cultivate resources in response to stressors and adverse structural conditions. Future studies should continue to investigate the complex set of factors which shape the mental well-being and aging of the U.S. population.

Cognitive impairment is a common non-motor symptom of Parkinson's disease (PD), yet the mechanism remains unclear.

Longitudinal serum samples from PD patients in a community-based case-control study were profiled, with cognition assessed by the Mini-Mental State Examination at baseline and follow-up. We performed three metabolome-wide association studies: a baseline cross-sectional analysis (n = 632), a time-to-event analysis (n = 412), and a repeated measures analysis (n = 265), followed by "Mummichog" pathway analysis.

Of the 4762 metabolomic features, four reached statistical significance in the baseline analysis, 22 in the time-to-event analysis, and none in the repeated measures analysis. Dysregulated pathways included purine metabolism, the pentose phosphate pathway, fatty acid pathways, and C21-steroid hormone biosynthesis and metabolism.

Disruptions in core metabolic pathways may contribute to PD-related cognitive impairments. Disturbances in pentose and purine pathways may occur before clinical cognitive decline is observed, suggesting early metabolic markers.

Attention-deficit hyperactivity disorder (ADHD) and conduct disorder (CD) are diagnoses that are frequently comorbid. CD is characterised by persistent dissocial, aggressive, and defiant behaviors that defy age-appropriate expectations. This is the case of a 15-year-old male adoptee referred to Tier 3 Child and Adolescent Mental Health Service (CAMHS) for severe aggression, impulsivity, and domestic violence. The patient exhibited coercive control toward his mother, physical violence involving weapons, and property damage, necessitating police involvement and risking family placement breakdown. Following a failure of psychosocial interventions, the patient underwent the Quantified Behavioral (Qb) Test and Conners-4 assessments and review by a consultant psychiatrist confirming ADHD and conduct disorder diagnosis. Methylphenidate was initiated and titrated to 36mg daily. Outcomes were significant. Within one month, antisocial behaviors reduced markedly. At seven months, domestic violence had ceased, and social care involvement ended. The patient successfully completed his General Certificates of Secondary Education (GCSE). Repeat testing on medication showed normalisation of scores: QbTest activity dropped to the 23rd percentile and impulsivity to the 21st percentile. Crucially, parent-rated Conners-4 scores for CD dropped to zero. This case highlights the importance of identifying and treating underlying ADHD in adolescents presenting with severe conduct issues. Higher doses of methylphenidate are often cited as beneficial for CD symptom reduction; this patient achieved remission of aggression at 36mg/day. The case shows that delaying ADHD diagnosis in favor of solely managing behavioral conduct symptoms can risk education and family stability. Treating the ADHD effectively mitigated the emotional dysregulation and violence and reduced the negative impact on this patient during a crucial period of his life as he was completing high school. This case highlights the importance of considering comorbid ADHD and CD in patients presenting with impulsive behaviours and aggression as their primary presenting complaints. Appropriate management with methylphenidate could have a significant therapeutic effect on these patients and their antisocial behaviours.

Developmental delay is a major pediatric concern, and children born preterm or requiring neonatal intensive care are at particularly increased risk. The Developmental Assessment Scale for Indian Infants (DASII) is a validated tool for evaluating motor and mental development in children up to 30 months of age, but detailed profiles of children referred for suspected developmental delay in Indian clinical settings remain sparse. This study aimed primarily to describe the developmental profile of children referred for developmental assessment using the DASII. As a secondary exploratory analysis, developmental measures were compared between term and preterm children, with recognition of the unstratified gestational grouping and potential differences in referral pathways and clinical characteristics.

This retrospective cross-sectional study analyzed medical records of 202 children aged 3-30 months who underwent DASII assessment between January 2023 and December 2025 at a developmental assessment center in Chhatrapati Sambhajinagar, Maharashtra, India. Motor and mental developmental quotients (Developmental Motor Quotient (DMoQ) and Developmental Mental Quotient (DMeQ)) and developmental ages were recorded. Outcomes were compared between term and preterm children using Student's t-test/Mann-Whitney U test and chi-square/Fisher's exact test, as appropriate.

Among 202 children assessed using DASII, motor developmental delay was present in 90.6% and mental developmental delay in 92.1%, with combined motor and mental delay being the predominant pattern (86.1%). In the exploratory term-preterm analysis, selected developmental scores and developmental ages differed between groups; however, DMoQ and DMeQ did not differ significantly after adjustment for multiple comparisons. A higher proportion of preterm children had no categorical mental developmental delay than term children; however, this finding should be interpreted cautiously given the referral-based design, unstratified gestational grouping, and small preterm subgroup.

Children referred for developmental assessment demonstrated a high burden of developmental impairment, with combined motor and mental delay being the predominant pattern. Differences between term and preterm children varied depending on the developmental measure used and should therefore be interpreted with caution. Prospective studies with larger sample sizes and comprehensive domain-specific assessments are warranted.

Epilepsy is a chronic neurologic condition that requires specialist follow-up and frequent medication adjustments. A multidisciplinary Veterans Affairs (VA) clinic model was implemented in which clinical pharmacist practitioners (CPPs) provide follow-up care under epileptologist oversight. This quality improvement study evaluated the economic and clinical impact of CPP-led follow-up visits in a VA epilepsy clinic, focusing on visit-level cost avoidance compared with neurologist-provided care.

This single-center, retrospective quality improvement project at a VA epilepsy clinic evaluated pharmacist-led follow-up epilepsy care between October 2021 and April 2025. The primary outcome was visit-level cost avoidance associated with CPP-led follow-up visits compared with physician provided follow-up. Cost avoidance was estimated using VA salary data, with each follow-up visit requiring approximately 1.5 h of work (30 min preparation, 30 min appointment, 30 min documentation), totaling 264 annual hours (12.7% full-time equivalent) at an annualized volume of 176 CPP visits. Secondary outcomes included reimbursement-equivalent value estimated through intervention-based assignment of evaluation and management (E/M) codes (99212-99,215) and 2025 Medicare Physician Fee Schedule rates, with a commercial payer scenario using a 1.3 multiplier, as well as clinical intervention patterns including the proportion of follow-up visits managed by CPPs versus physicians, frequencies of antiseizure medication (ASM) changes, nonpharmacologic interventions, mental health screenings, referrals, laboratory orders, and mean wait time for follow-up appointment. New patient visits, nurse practitioner visits, and non-epilepsy neurology encounters were excluded from the cost analysis.

Over 3.5 years, 1596 total encounters were completed, including 1214 followup visits analyzed for cost and clinical outcomes. Of these follow-up visits, CPPs managed 617 (51%) and neurologists 597 (49%). Intervention-based EM classification yielded a base-case distribution of 10% 99,212, 55% 99,213, 30% 99,214, and 5% 99,215, corresponding to an annualized Medicare-equivalent value of 17,478 USD (United States Dollar), increasing to 22,722 USD with the commercial multiplier. VA salary modeling estimated to be 114.80 USD per CPP follow-up visit versus 216.50 USD per neurologist visit, corresponding to 101 USD cost avoidance per CPP encounter and approximately 17,888 USD annual cost avoidance at an annualized volume of 176 CPP follow-up visits.Mean follow-up wait time decreased from 29 to 8 days after CPP integration. Among CPP encounters, ASM changes occurred in 236 of 617 visits (38%), nonpharmacologic interventions in 89 visits (14%), and mental health screenings in 325 visits (53%), with CPPs also placing 44 specialty consultations (7%) and ordering 146 laboratory tests (24%). Peer review of 70 CPP progress notes in epilepsy clinic between 2022 and 2025 found all met predefined criteria for accuracy and clinical appropriateness.

CPP integration into a VA epilepsy clinic enabled pharmacists to absorb over half of follow-up volume while shortening wait times and delivering clinically meaningful care. Pharmacist-led encounters were characterized by frequent ASM adjustments, mental health screening, and safety and adherence counseling and produced substantial visit-level cost avoidance and demonstrable reimbursement equivalent value compared with neurologist staffing. These findings support CPP-led follow-up as a strategy to expand access, preserve epileptologist capacity, and reduce per-visit costs.

The Collaborative Care Model (CoCM), which integrates mental health services into primary care through a team-based approach involving GPs, care managers (CM), and mental health consultants, is a proven way to meet these challenges. A first CoCM adaptation in French primary care settings called SESAME has been implemented since 2021.

Semi-structured interviews were conducted with all the healthcare professionals involved in SESAME to examine the facilitators and barriers experienced during the first year. Data were analyzed using a qualitative approach combining manual thematic analysis and lexicometric analysis with ALCESTE software.

Four discourse classes emerged from the analysis. Key facilitators included prior collaborative experience, effective digital communication tools, and clear role understanding. Barriers such as technological challenges, role ambiguity, and referral framework misunderstandings were also identified. The structured inclusion process and rapid access to care were appreciated, but challenges persisted in referral processes and role boundary definitions. Financial coverage and a less stigmatizing approach were seen as benefits.

This study provides insights into the early implementation dynamics of CoCM in a novel context, underscoring the need for clear protocols, ongoing communication, and tailored support to ensure successful integration of mental health services into primary care.

There can be no assurance about the mental health of college students in today's digital information age without their personal digital health literacy. The current academic studies on this mechanism are in their infancy. This paper uses cognitive behavioral theory and stress response theory to build a chain mediation model of internet addiction and sleep quality. The study recruited participants from a pool of 407 college students and provided the students with standard scales, such as the e-HEALS, AMHLAQ, IAT, and ISI-C. The researchers used the SPSS 27.0 software to analyze the data and tested the mediation hypothesis using the Bootstrap method. The findings indicate a positive relationship between digital health literacy and mental health among college students. Additionally, the study found that internet addiction and sleep quality served as mediators. The complete chain was defined as digital health literacy → internet addiction → sleep quality, with the total indirect effect on college mental health measured at 0.6304. The study helped fill an academic void regarding digital health literacy. It demonstrated that internet addiction can be reduced and sleep quality and college students' mental health can be improved. The study has both academic and professional merit. The study aids colleges in the application of digital intervention strategies and the development of mental health literacy among college students.

Digital media's mental-health signature may depend on which platform a population uses, yet nationally representative evidence that distinguishes platforms is scarce, especially for upper-middle-income, Muslim-majority, non-Arab settings. We examined platform-specific associations between digital media use and psychological distress in the Turkish adult population, and whether age moderates them, using the 2024 Turkish General Social Survey, a probability-based, nationally representative survey of adults aged 18 and over (N = 2,615; analytical N = 1,478 after listwise deletion). Psychological distress was measured with a nine-item past-week negative-affect index informed by the Center for Epidemiologic Studies Depression Scale (CES-D), and unweighted hierarchical ordinary least squares regression with heteroskedasticity-consistent (HC3) robust standard errors modeled demographic, aggregate-exposure, platform-specific, age-interaction, and loneliness blocks, with a Johnson-Neyman procedure characterizing age moderation. Instagram use (β = +.066, p = .011) was associated with higher psychological distress and WhatsApp use (β = -.073, p = .007) with lower distress, whereas total internet time, TikTok, and YouTube showed no detectable association. Overall social-media time was likewise positively associated with distress (β = +.088, p = .016), but the platform contrasts, not aggregate exposure, survived the most demanding clustering and exposure-scale checks. The positive Instagram association declined with age and was statistically indistinguishable from zero by the early forties, a secondary finding that turned on whether loneliness entered the model (age-by-Instagram β = -.069, p = .014 with loneliness, β = -.060, p = .059 without). Adding loneliness - an affective covariate overlapping the outcome battery - raised explained variance to R ² = .359 (largely shared variance) while the platform and aggregate coefficients remained significant. In this national sample the platform types diverged: algorithm-driven image-platform use (Instagram) showed a small positive association with distress, interpersonal-messaging use (WhatsApp) a small inverse (protective-direction) one. The headline effect sizes were below Cohen's small-effect benchmark and estimated at the limit of what this design could detect, and the cross-sectional data require longitudinal confirmation of the platform contrasts. Whether distinguishing platform types, rather than treating screen time as uniform, improves population-level digital well-being efforts remains a hypothesis for longitudinal and experimental testing.