{"title":"Speech Emotion Recognition in Mental Health: Systematic Review of Voice-Based Applications.","authors":"Eric Jordan, Raphaël Terrisse, Valeria Lucarini, Motasem Alrahabi, Marie-Odile Krebs, Julien Desclés, Christophe Lemey","doi":"10.2196/74260","DOIUrl":"https://doi.org/10.2196/74260","url":null,"abstract":"<p><strong>Background: </strong>The field of speech emotion recognition (SER) encompasses a wide variety of approaches, with artificial intelligence technologies providing improvements in recent years. In the domain of mental health, the links between individuals' emotional states and pathological diagnoses are of particular interest.</p><p><strong>Objective: </strong>This study aimed to investigate the performance of tools combining SER and artificial intelligence approaches with a view to their use within clinical contexts and to determine the extent to which SER technologies have already been applied within clinical contexts.</p><p><strong>Methods: </strong>The review includes studies applied to speech (audio) signals for a select set of pathologies or disorders and only includes those studies that evaluate diagnostic performance using machine learning performance metrics or statistical correlation measures. The PubMed, IEEE Xplore, arXiv, and ScienceDirect databases were queried as recently as February 2025. The Quality Assessment of Diagnostic Accuracy Studies tool was used to measure the risk of bias.</p><p><strong>Results: </strong>A total of 14 articles were included in the final review. The included papers addressed suicide risk (3/14, 21%), depression (8/14, 57%), and psychotic disorders (3/14, 21%).</p><p><strong>Conclusions: </strong>SER technologies are mostly used indirectly in mental health research and in a wide variety of ways, including different architectures, datasets, and pathologies. This diversity makes a direct assessment of the technology challenging. Nonetheless, promising results are obtained in various studies that attempt to diagnose patients based on either indirect or direct results from SER models. These results highlight the potential for this technology to be used within a clinical setting. Future work should focus on how clinicians can use these technologies collaboratively.</p><p><strong>Trial registration: </strong>PROSPERO CRD420251006669; https://www.crd.york.ac.uk/PROSPERO/view/CRD420251006669.</p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e74260"},"PeriodicalIF":5.8,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145201665","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Smartphone-Based Mindfulness and Mentalization Ecological Momentary Interventions for Common Mental Health Problems: Pilot Randomized Controlled Trial.","authors":"Ciarán O'Driscoll, Sarah O'Reilly, Agata Julita Jaremba, Tobias Nolte, Madiha Shaikh","doi":"10.2196/79296","DOIUrl":"10.2196/79296","url":null,"abstract":"<p><strong>Background: </strong>Accessible ecological momentary interventions deliver brief, real-time support integrated into daily routines. Interpersonal dynamics and maladaptive coping mechanisms can contribute to an individual's anxiety and depression. Both mindfulness and mentalization represent psychological constructs with the potential to mitigate the negative impact of interpersonal stressors.</p><p><strong>Objective: </strong>This study aims to assess the feasibility and acceptability of an automated mindfulness- and mentalization-based ecological momentary intervention for common mental health problems as delivered via a mobile phone app.</p><p><strong>Methods: </strong>The design was a parallel-group pilot randomized controlled trial with 1:1 allocation ratio and exploratory framework. Recruitment of participants experiencing common mental health issues was internet-based from a university setting. Eligible participants were randomly allocated to fully automated mindfulness- or mentalization-based ecological momentary interventions via computer-generated randomization. Participants were blind to the alternative intervention options. Outcomes were self-assessed through questionnaires after 4 weeks. Primary outcomes were feasibility (recruitment, retention, and adherence) and acceptability (satisfaction ratings and qualitative feedback). Secondary outcomes included changes in depression (Patient Health Questionnaire-9 [PHQ-9]) and anxiety (Generalized Anxiety Disorder Questionnaire-7 [GAD-7]) scores.</p><p><strong>Results: </strong>A total of 84 participants were randomized (42 to each group). The interventions demonstrated good feasibility with an 89.2% retention rate and a mean adherence of 87.69% (SD 11.3%) across both groups. Acceptability ratings were positive, with favorable scores for ease of engagement (mean 5.20, SD 1.6), overall enjoyment (mean 5.15, SD 1.2), and likelihood of recommending the app (mean 5.11, SD 1.6) on a 7-point scale. For primary outcomes, both groups showed significant within-group reductions in PHQ-9 and GAD-7 scores, with moderate to large effect sizes (Cohen d=-0.68 to -0.81), with no significant difference between groups. Both treatments demonstrated clinically significant change, with 33 (44%) participants in both groups no longer meeting caseness criteria for anxiety and depression. Mindfulness performed better on improving assertiveness and perceived support compared to mentalization in the ecological momentary assessment data. One unintended harm was reported in the mindfulness arm, whereas none was reported in the mentalization arm.</p><p><strong>Conclusions: </strong>This pilot trial suggests that both mindfulness- and mentalization-based ecological momentary interventions are feasible and acceptable for individuals with common mental health problems and warrant further evaluation.</p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e79296"},"PeriodicalIF":5.8,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12483342/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145201651","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Brittany Quinn, Lindsey Nichols, Jennifer Frazee, Mark Payton, Rachel M A Linger
{"title":"Dissemination of Information on Selective Serotonin Reuptake Inhibitors on TikTok: Analytical Mixed Methods Study of Creator Types, Content Tone, and User Engagement.","authors":"Brittany Quinn, Lindsey Nichols, Jennifer Frazee, Mark Payton, Rachel M A Linger","doi":"10.2196/77383","DOIUrl":"10.2196/77383","url":null,"abstract":"<p><strong>Background: </strong>TikTok [ByteDance] is a significant source of mental health-related content, including discussions on selective serotonin reuptake inhibitors (SSRIs). While the app fosters community building, its algorithm also amplifies misinformation as influencers without relevant expertise often dominate conversations about SSRIs. These videos frequently highlight personal experiences, potentially overshadowing evidence-based information from health care professionals. Despite these concerns, TikTok holds potential as a tool for improving mental health literacy when used by professionals to provide credible information.</p><p><strong>Objective: </strong>This study aimed to examine TikTok videos on SSRIs, hypothesizing that content will predominantly emphasize negative experiences and that videos by nonmedical professionals will attract higher engagement. By analyzing creators, engagement metrics, content tone, and video tone, this study aimed to shed light on social media's role in shaping perceptions of SSRIs and mental health literacy.</p><p><strong>Methods: </strong>A sample of 99 TikTok videos was collected on December 8, 2024. Apify, a web scraper, compiled pertinent engagement metrics (URLs, likes, comments, and shares). Views were manually recorded. In total, 3 researchers evaluated video and content tones and documented findings in Qualtrics. User profiles were analyzed to classify creators as a \"medical professional\" or \"nonmedical professional\" based on verification of their credentials. Statistical analyses evaluated the hypotheses.</p><p><strong>Results: </strong>The number of videos created by both nonmedical and medical professionals was roughly even. Approximately one-third (35/99, 35%) mentioned a specific SSRI (ie, fluoxetine, fluvoxamine, vilazodone, sertraline, paroxetine, citalopram, or escitalopram). Compared to medical professionals, nonmedical creators produced significantly more videos with a positive video tone (P<.001). TikToks made by both groups of creators, however, had negative content tones (P=.78). Nonmedical professionals received significantly greater overall views (P=.01), likes (P=.01), and comments (P=.03), but overall shares were not significantly different (P=.18). Daily interaction metrics revealed that nonmedical professionals received more daily interaction, but these differences were not significant in terms of views (P=.09), likes (P=.06), comments (P=.15), or shares (P=.28).</p><p><strong>Conclusions: </strong>Results showed that while both creator groups focused on negative SSRI side effects and experiences (content tone), the way they presented this information (video tone) differed. Medical professionals generally maintained a neutral video tone, whereas nonmedical professionals were more likely to adopt a positive video tone. This may explain why nonmedical professionals' videos had significantly more cumulative views, likes, and comments than medical professionals' videos. Thes","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e77383"},"PeriodicalIF":5.8,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12483472/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145201546","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A Bibliometric and Visual Analysis of Artificial Intelligence Applications in Depression Detection and Diagnosis: Trends and Future Directions.","authors":"Wenbo Ren, Xiali Xue, Lu Liu, Jiahuan Huang","doi":"10.2196/79293","DOIUrl":"https://doi.org/10.2196/79293","url":null,"abstract":"<p><strong>Background: </strong>Depression is a highly prevalent and debilitating mental disorder, yet its diagnosis heavily relies on subjective assessments, leading to challenges in accuracy and consistency. The rapid advancements in Artificial Intelligence (AI) offer promising avenues for more objective and efficient diagnostic approaches. Understanding the evolving landscape of AI applications in depression diagnosis is crucial for guiding future research and clinical translation.</p><p><strong>Objective: </strong>This study aims to provide a comprehensive bibliometric and visual analysis of the global research trends, intellectual structure, and emerging frontiers in the application of AI for depression detection and diagnosis from 2015 to 2024.</p><p><strong>Methods: </strong>A systematic literature search was conducted on the Web of Science Core Collection (WoSCC) database to identify relevant publications on AI applications in depression diagnosis from January 1, 2015, to December 31, 2024. A total of 2304 articles were retrieved and analyzed using bibliometric software CiteSpace. The analysis encompassed temporal trends, keyword dynamics, author collaboration networks, institutional influence, country contributions, and intellectual foundations through co-citation analysis of journals and references.</p><p><strong>Results: </strong>The field exhibited exponential growth in publications and citations, particularly after 2018, reflecting increasing academic and clinical interest. Key thematic shifts were observed from traditional machine learning to advanced deep learning, multimodal fusion, and the integration of objective biomarkers (e.g., EEG, facial expressions). Leading contributors included institutions from China and the United States, with forming collaborative bridges from countries like Canada and Singapore. The intellectual base is highly interdisciplinary, drawing heavily from computer science, neuroscience, and psychiatry, with a notable surge in engineering and translational research.</p><p><strong>Conclusions: </strong>The integration of AI in depression diagnosis is a rapidly maturing and diversifying field, transitioning from theoretical exploration to clinically relevant applications focusing on objective, data-driven approaches. The identified trends underscore the need for enhanced interdisciplinary and international collaboration, ethical framework development, and a focus on translating technological innovations into accessible and equitable mental health solutions. These findings offer valuable insights for researchers, clinicians, and policymakers to strategically advance AI-assisted depression diagnostics globally.</p><p><strong>Clinicaltrial: </strong></p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":" ","pages":""},"PeriodicalIF":5.8,"publicationDate":"2025-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145193556","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Heather Robinson, Millissa Booth, Lauren Fothergill, Claire Friedrich, Zoe Glossop, Jade Haines, Andrew Harding, Rose Johnston, Steven Jones, Karen Machin, Rachel Meacock, Kristi Nielson, Paul Marshall, Jo-Anne Puddephatt, Tamara Rakić, Paul Rayson, Jo Rycroft-Malone, Nick Shryane, Zoe Swithenbank, Sara Wise, Fiona Lobban
{"title":"Understanding the Needs of Moderators in Online Mental Health Forums: Realist Synthesis and Recommendations for Support.","authors":"Heather Robinson, Millissa Booth, Lauren Fothergill, Claire Friedrich, Zoe Glossop, Jade Haines, Andrew Harding, Rose Johnston, Steven Jones, Karen Machin, Rachel Meacock, Kristi Nielson, Paul Marshall, Jo-Anne Puddephatt, Tamara Rakić, Paul Rayson, Jo Rycroft-Malone, Nick Shryane, Zoe Swithenbank, Sara Wise, Fiona Lobban","doi":"10.2196/58891","DOIUrl":"https://doi.org/10.2196/58891","url":null,"abstract":"<p><strong>Background: </strong>There has been an increase in the use of online mental health forums to support mental health. These forums are often moderated by trained moderators to ensure a safe, therapeutic environment. While the moderator role is rewarding, it can also be challenging. There is a need to understand the impact of the role on moderators and how they can best be supported to maintain psychological well-being.</p><p><strong>Objective: </strong>This study aimed to understand how, why, and in what contexts moderator well-being is affected by the moderator role and produce actionable recommendations for how moderators can best be supported to maintain workplace psychological well-being.</p><p><strong>Methods: </strong>We conducted realist synthesis of (1) published and gray literature from 2019 to 2023, (2) stakeholder interviews with forum moderators and hosts, and (3) moderator training manuals developed by organizations that host online mental health forums. Self-determination theory was used as the theoretical basis for this synthesis.</p><p><strong>Results: </strong>We developed 24 context-mechanism-outcome configurations from our realist analysis of 9 published papers, 18 interviews, and 5 training manuals. The findings highlight the specific ways in which moderator well-being can be supported through meeting the psychological needs for autonomy, competence, and relatedness. Forums that allow moderators to work in alignment with their personal motivations can increase moderator well-being. Forum organizations should support moderator competence through initial expectation setting, especially around moderator responsibility for user well-being, and ongoing support, such as meaningful supervision and peer support. Co-designed training, reflective practice, and experiential learning are all key to increasing moderator competence and satisfaction in the workplace. Working within a diverse team with access to innovative forum design can increase moderator psychological well-being. Organizational support for moderators' well-being through monitoring and encouraging self-care is vital to ensure moderators can effectively carry out their role. Making and supporting meaningful relationships in the forum can boost psychological well-being and the therapeutic value of the moderator role. Key challenges for moderators were dealing with conflicts between supporting open discussion and ensuring a safe community environment, sharing lived experiences in positive ways for both moderator and user, and supporting people within the limitations of an anonymous forum.</p><p><strong>Conclusions: </strong>This realist synthesis is the first to examine the impacts on well-being of being a moderator of an online mental health forum. Recommendations to support moderator psychological well-being are proposed, targeted at specific stakeholder groups to aid implementation. Organizational-level endorsement and facilitation of support are particularly impo","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e58891"},"PeriodicalIF":5.8,"publicationDate":"2025-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145179377","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A Prompt Engineering Framework for Large Language Model-Based Mental Health Chatbots: Design Principles and Insights for AI-Supported Care.","authors":"Sorio Boit, Rajvardhan Patil","doi":"10.2196/75078","DOIUrl":"https://doi.org/10.2196/75078","url":null,"abstract":"<p><strong>Background: </strong>Artificial intelligence (AI), particularly large language models (LLMs), presents a significant opportunity to transform mental healthcare through scalable, on-demand support. While LLM-powered chatbots may help reduce barriers to care, their integration into clinical settings raises critical concerns regarding safety, reliability, and ethical oversight. A structured framework is needed to capture their benefits while addressing inherent risks. This paper introduces a conceptual model for prompt engineering, outlining core design principles for the responsible development of LLM-based mental health chatbots.</p><p><strong>Objective: </strong>This paper proposes a comprehensive, layered framework for prompt engineering that integrates evidence-based therapeutic models, adaptive technology, and ethical safeguards. The objective is to propose and outline a practical foundation for developing AI-driven mental health interventions that are safe, effective, and clinically relevant.</p><p><strong>Methods: </strong>We outline a layered architecture for an LLM-based mental health chatbot. The design incorporates: (1) an input layer with proactive risk detection; (2) a dialogue engine featuring a user state database for personalization and Retrieval-Augmented Generation (RAG) to ground responses in evidence-based therapies such as Cognitive Behavioral Therapy (CBT), Acceptance and Commitment Therapy (ACT), and Dialectical Behavior Therapy (DBT); and (3) a multi-tiered safety system, including a post-generation ethical filter and a continuous learning loop with therapist oversight.</p><p><strong>Results: </strong>The primary contribution is the framework itself, which systematically embeds clinical principles and ethical safeguards into system design. We also propose a comparative validation strategy to evaluate the framework's added value against a baseline model. Its components are explicitly mapped to the FAITA-MH and READI frameworks, ensuring alignment with current scholarly standards for responsible AI development.</p><p><strong>Conclusions: </strong>The framework offers a practical foundation for the responsible development of LLM-based mental health support. By outlining a layered architecture and aligning it with established evaluation standards, this work offers guidance for developing AI tools that are technically capable, safe, effective, and ethically sound. Future research should prioritize empirical validation of the framework through the phased, comparative approach introduced in this paper.</p><p><strong>Clinicaltrial: </strong></p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":" ","pages":""},"PeriodicalIF":5.8,"publicationDate":"2025-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145151129","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Zhao Hui Koh, Duygu Serbetci, Jason Skues, Greg Murray
{"title":"Toward Digital Self-Monitoring of Mental Health in the General Population: Scoping Review of Existing Approaches to Self-Report Measurement.","authors":"Zhao Hui Koh, Duygu Serbetci, Jason Skues, Greg Murray","doi":"10.2196/59351","DOIUrl":"10.2196/59351","url":null,"abstract":"<p><strong>Background: </strong>With the ubiquity of smartphones, digital self-report instruments have enormous potential to support the general population in monitoring their mental health. A primary challenge for researchers committed to advancing this work is simply to scope the plethora of widely used candidate instruments. The overarching aim of this study was to address this challenge to support and guide future research in this burgeoning area.</p><p><strong>Objective: </strong>This study aimed to conduct a literature review of self-report instruments used in empirical studies to measure mental health (1) in the general population, (2) delivered in a digital format, and (3) in longitudinal designs. Given the wide range of recognized \"mental health\" constructs, the review's search strategies were guided by Keyes' dual continua model of mental health, recognizing both deficits- and strengths-based constructs. This study's primary objective was to develop a first-of-its-kind ranking and synthesis of the most frequently used instruments that are potentially suitable for mental health self-monitoring. It was not an objective of this study to evaluate psychometric properties of the identified instruments-we hope the present ranking and synthesis will provide the foundation for future research into optimal digital, prospective self-report of mental health.</p><p><strong>Methods: </strong>Five major electronic databases were searched. Studies that administered digital mental health instruments (in English) repeatedly to community dwellers in the general adult population were eligible. The included studies were grouped by instruments for synthesis using a narrative approach.</p><p><strong>Results: </strong>Preliminary screening of 95,849 records identified 8460 eligible records, among which 1000 records were randomly selected over 4 iterations for full-text screening. A total of 223 records were included. We found that the top 30 most commonly used instruments accounted for 78.4% (308/393) of the total usage across studies. These instruments predominantly measure deficits-based mental health constructs. The Patient Health Questionnaire 9 Items and Generalized Anxiety Disorder 7 Items were by far the most used instruments. The most commonly measured strengths-based constructs were life satisfaction and mental well-being.</p><p><strong>Conclusions: </strong>The findings of this review strongly suggest that scientific investigation of mental health constructs across time on digital platforms still prioritizes deficits-focused instruments originally developed for pen-and-paper administration using classical test theory. These findings are discussed in light of evidence in the literature that deficits-focused instruments demonstrate inferior distributional properties (floor effects) in the general population and theory suggesting that both deficits- and strengths-focused measurements are required to holistically assess mental health. Limitations of the ","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e59351"},"PeriodicalIF":5.8,"publicationDate":"2025-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145087733","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Boyu Ren, WonJin Yoon, Spencer Thomas, Guergana Savova, Timothy Miller, Mei-Hua Hall
{"title":"Cross-Site Predictions of Readmission After Psychiatric Hospitalization With Mood or Psychotic Disorders: Retrospective Study.","authors":"Boyu Ren, WonJin Yoon, Spencer Thomas, Guergana Savova, Timothy Miller, Mei-Hua Hall","doi":"10.2196/71630","DOIUrl":"10.2196/71630","url":null,"abstract":"<p><strong>Background: </strong>Patients with mood or psychotic disorders experience high rates of unplanned hospital readmissions. Predicting the likelihood of readmission can guide discharge decisions and optimize patient care.</p><p><strong>Objective: </strong>The purpose of this study is to evaluate the predictive power of structured variables from electronic health records for all-cause readmission across multiple sites within the Mass General Brigham health system and to assess the transportability of prediction models between sites.</p><p><strong>Methods: </strong>This retrospective, multisite study analyzed structured variables from electronic health records separately for each site to develop in-site prediction models. The transportability of these models was evaluated by applying them across different sites. Predictive performance was measured using the F1-score, and additional adjustments were made to account for differences in predictor distributions.</p><p><strong>Results: </strong>The study found that the relevant predictors of readmission varied significantly across sites. For instance, length of stay was a strong predictor at only 3 of the 4 sites. In-site prediction models achieved an average F1-score of 0.661, whereas cross-site predictions resulted in a lower average F1-score of 0.616. Efforts to improve transportability by adjusting for differences in predictor distributions did not improve performance.</p><p><strong>Conclusions: </strong>The findings indicate that individual site-specific models are necessary to achieve reliable prediction accuracy. Furthermore, the results suggest that the current set of predictors may be insufficient for cross-site model transportability, highlighting the need for more advanced predictor variables and predictive algorithms to gain robust insights into the factors influencing early psychiatric readmissions.</p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e71630"},"PeriodicalIF":5.8,"publicationDate":"2025-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12431164/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145056053","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Allyson Cruickshank, Pantelis Andreou, Debbie Johnson Emberly, Sandra Meier, Leslie Anne Campbell
{"title":"Child and Adolescent Virtual Mental Health Care and Duration of Treatment: Retrospective Cohort Study.","authors":"Allyson Cruickshank, Pantelis Andreou, Debbie Johnson Emberly, Sandra Meier, Leslie Anne Campbell","doi":"10.2196/70650","DOIUrl":"10.2196/70650","url":null,"abstract":"<p><strong>Background: </strong>Due to public health restrictions, the COVID-19 pandemic required significant changes in the delivery of child and adolescent mental health services. The use of virtual care for balancing access with treatment needs requires a shared decision between clients, caregivers, and clinicians. One aspect for consideration is the length of treatment necessary to achieve desired outcomes and whether it differs by treatment modality. Insights gained from the comparison of treatment duration between modalities may improve our understanding of the effectiveness of virtual care and help to inform clinical decision-making and effective use of resources.</p><p><strong>Objective: </strong>We sought to improve our understanding of how treatment modality impacts treatment duration for children and adolescents accessing Community Mental Health and Addictions services at IWK Health following the rapid implementation of virtual care in March 2020. In this study, we aimed to compare the duration of treatment within episodes of care by treatment modality and determine whether client characteristics, system factors, or time period influenced any associations between treatment modality and treatment duration.</p><p><strong>Methods: </strong>Episodes of care were created using administrative data collected by the IWK Mental Health and Addictions program and used as the unit of analysis. A multilevel mixed-effects negative binomial model and time-to-event analysis were used to model the association between treatment modality and treatment duration, both in visits and days, adjusting for client and system characteristics.</p><p><strong>Results: </strong>Virtual episodes of care had more visits than in-person episodes between April 1, 2020, and March 31, 2021 (incidence rate ratio [IRR] 1.59, 95% CI 1.38-1.83), and April 1, 2021, and March 31, 2022 (IRR 1.22, 95% CI 1.10-1.35), whereas between April 1, 2022, and March 31, 2023, virtual episodes of care were associated with fewer visits (IRR 0.82, 95% CI 0.74-0.91). Comparable results were seen for treatment duration in days (2020-2021: hazard ratio [HR] 0.64, 95% CI 0.54-0.76; 2021-2022: HR 0.80, 95% CI 0.70-0.90; and 2022-2023: HR 1.10, 95% CI 0.97-1.25). These differences by time period relative to the onset of the COVID-19 pandemic and switch to virtual care were consistent after adjusting for client and system characteristics.</p><p><strong>Conclusions: </strong>To our knowledge, this is the first study to examine the association between virtual or in-person treatment modality and treatment duration. While initially longer than in-person episodes of care, both in numbers of visits and length in days, over time the average length of episodes conducted mainly virtually had attenuated. These findings may be due to growing comfort with the technology or client factors not adequately captured in administrative data. This information can be valuable to clinicians, clients, and their families reg","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e70650"},"PeriodicalIF":5.8,"publicationDate":"2025-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12425424/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145041966","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Iftikhar Ahmed, Anushree Brahmacharimayum, Raja Hashim Ali, Talha Ali Khan, Muhammad Ovais Ahmad
{"title":"Explainable AI for Depression Detection and Severity Classification From Activity Data: Development and Evaluation Study of an Interpretable Framework.","authors":"Iftikhar Ahmed, Anushree Brahmacharimayum, Raja Hashim Ali, Talha Ali Khan, Muhammad Ovais Ahmad","doi":"10.2196/72038","DOIUrl":"10.2196/72038","url":null,"abstract":"<p><strong>Background: </strong>Depression is one of the most prevalent mental health disorders globally, affecting approximately 280 million people and frequently going undiagnosed or misdiagnosed. The growing ubiquity of wearable devices enables continuous monitoring of activity levels, providing a new avenue for data-driven detection and severity assessment of depression. However, existing machine learning models often exhibit lower performance when distinguishing overlapping subtypes of depression and frequently lack explainability, an essential component for clinical acceptance.</p><p><strong>Objective: </strong>This study aimed to develop and evaluate an interpretable machine learning framework for detecting depression and classifying its severity using wearable-actigraphy data, while addressing common challenges such as imbalanced datasets and limited model transparency.</p><p><strong>Methods: </strong>We used the Depresjon dataset and applied Adaptive Synthetic Sampling (ADASYN) to mitigate class imbalance. We extracted multiple statistical features (eg, power spectral density mean and autocorrelation) and demographic attributes (eg, age) from the raw activity data. Five machine learning algorithms (logistic regression, support vector machines, random forest, XGBoost, and neural networks) were assessed via accuracy, precision, recall, F1-score, specificity, and Matthew correlation constant. We further used Shapley Additive Explanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME) to elucidate prediction drivers.</p><p><strong>Results: </strong>XGBoost achieved the highest overall accuracy of 84.94% for binary classification and 85.91% for multiclass severity. SHAP and LIME revealed power spectral density mean, age, and autocorrelation as top predictors, highlighting circadian disruptions' role in depression.</p><p><strong>Conclusions: </strong>Our interpretable framework reliably identifies depressed versus nondepressed individuals and differentiates mild from moderate depression. The inclusion of SHAP and LIME provides transparent, clinically meaningful insights, emphasizing the potential of explainable artificial intelligence to enhance early detection and intervention strategies in mental health care.</p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e72038"},"PeriodicalIF":5.8,"publicationDate":"2025-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12425426/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145041974","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}